Post-Training Quantization enables efficient Vision Transformer (ViTs) deployment with a small calibration data, and its prevalent use of uniform quantization harnesses AI accelerator matrix cores for high-speed inference. However, the application of uniform quantization is fundamentally challenged by the extreme non-uniformity of activation distributions.Specifically, the power-law nature of post-Softmax attention scores and the significant inter-channel variance in post-GELU activations create a dilemma for conventional quantization, as it struggles to preserve critical high-magnitude values without sacrificing overall precision. To resolve this core conflict, we introduce UQ-ViT (Uniform Quantization for Vision Transformers), a novel uniform quantization framework designed to reconcile high precision with hardware efficiency. Central to UQ-ViT are two operators: Dynamic Elimination of Maximum (DeMax) and Normalization Quantization (NormQuant). DeMax is a quantization operator for post-Softmax attention scores that utilizes uniform quantization. It dynamically eliminates and preserves dominant values, effectively mitigating quantization loss from the extreme values in the power-law distribution. NormQuant utilizes a per-channel quantization strategy during quantization and reverts to a per-tensor format for dequantization, achieving both high accuracy and computational efficiency. Crucially, it is applicable to any linear layer, enabling effective quantization of post-GELU activations in ViTs. Through extensive experiments on various ViTs and vision tasks, including image classification, object detection, and instance segmentation, we demonstrate that our proposed approach outperforms existing methods, achieving superior accuracy while ensuring hardware friendliness.

Reference Audio-Visual Segmentation (Ref-AVS) tasks challenge models to precisely locate sounding objects by integrating visual, auditory, and textual cues. Existing methods often lack genuine semantic understanding, tending to memorize fixed reasoning patterns. Furthermore, jointly training for reasoning and segmentation can compromise pixel-level precision. To address these issues, we introduce AURORA, a novel framework designed to enhance genuine reasoning and language comprehension in reference audio-visual segmentation. We employ a structured Chain-of-Thought (CoT) prompting mechanism to guide the model through a step-by-step reasoning process and introduce a novel segmentation feature distillation loss to effectively integrate these reasoning abilities without sacrificing segmentation performance. To further cultivate the model's genuine reasoning capabilities, we devise a further two-stage training strategy: first, a ``corrective reflective-style training" stage utilizes self-correction to enhance the quality of reasoning paths, followed by reinforcement learning via Group Reward Policy Optimization (GRPO) to bolster robustness in challenging scenarios. Experiments demonstrate that AURORA achieves state-of-the-art performance on Ref-AVS benchmarks and generalizes effectively to unreferenced segmentation.

Few-shot semantic segmentation (FSS) often struggles with the intra-class diversity issue between query and support images, caused by the category-biased information provided by limited annotated support images for matching objects. While increasing the number of annotated support images could mitigate this bias, it is impractical within the few-shot learning framework. Therefore, our proposed Unlabeled Query Integration Few-Shot Segmentation (UQI-FSS) tackles this challenge by incorporating unlabeled query images into the learning paradigm. This approach aims to achieve a more comprehensive category representation, which is essential to enhance segmentation accuracy in various scenarios. However, integrating unlabeled query images directly requires careful management to prevent the dilution of vital information from the annotated support set. To address this issue, we present an Unlabeled Query Integration Network (UQINet), which adaptively extracts beneficial and suppresses detrimental information from the unlabeled query images. Specifically, we first introduce an Information Bridging Module to close the gap between support and unlabeled query features, generating a pseudo-support set enriched with additional category data. Next, we introduce a Query Fusion Module to incorporate query information from both prototype and pixel levels into the pseudo-support features, thus improving their adaptability to the query. Finally, we propose an Adaptive Selection Module to select effective category information and combine the pseudo-support features, thereby activating target objects for precise segmentation prediction. Experimental results show considerable performance improvements over previous methods on various FSS benchmarks. The application of this method to four challenging scenarios further underscores its versatility and practical value.

Salient object detection (SOD) and camouflaged object detection (COD) are related but distinct binary mapping tasks, each involving multiple modalities that share commonalities while maintaining unique characteristics. Existing approaches often rely on complex, task-specific architectures, leading to redundancy and limited generalization. Our previous work, VSCode, introduced a generalist model that effectively handles four SOD tasks and two COD tasks. VSCode leveraged VST as its foundation model and incorporated 2D prompts within an encoder-decoder framework to capture domain and task-specific knowledge, utilizing a prompt discrimination loss to optimize the model. Building upon the proven effectiveness of our previous work VSCode, we identify opportunities to further strengthen generalization capabilities through focused modifications in model design and optimization strategy. To unlock this potential, we propose VSCode-v2, an extension that introduces a Mixture of Prompt Experts (MoPE) layer to generate adaptive prompts. We also redesign the training process into a two-stage approach: first learning shared features across tasks, then capturing specific characteristics. To preserve knowledge during this process, we incorporate distillation from our conference version model. Furthermore, we propose a contrastive learning mechanism with data augmentation to strengthen the relationships between prompts and feature representations. VSCode-v2 demonstrates balanced performance improvements across six SOD and COD tasks. Moreover, VSCode-v2 effectively handles various multimodal inputs and exhibits zero-shot generalization capability to novel tasks, such as RGB-D Video SOD.

Low-light image enhancement, particularly in cross-domain tasks such as mapping from the raw domain to the sRGB domain, remains a significant challenge. Many deep learning-based methods have been developed to address this issue and have shown promising results in recent years. However, single-stage methods, which attempt to unify the complex mapping across both domains, leading to limited denoising performance. In contrast, existing two-stage approaches typically overlook the characteristic of demosaicing within the Image Signal Processing (ISP) pipeline, leading to color distortions under varying lighting conditions, especially in low-light scenarios. To address these issues, we propose a novel Mamba-based method customized for low light RAW images, called RAWMamba, to effectively handle raw images with different CFAs. Furthermore, we introduce a Retinex Decomposition Module (RDM) grounded in Retinex prior, which decouples illumination from reflectance to facilitate more effective denoising and automatic non-linear exposure correction, reducing the effect of manual linear illumination enhancement. By bridging demosaicing and denoising, better enhancement for low light RAW images is achieved. Experimental evaluations conducted on public datasets SID and MCR demonstrate that our proposed RAWMamba achieves state-of-the-art performance on cross-domain mapping. The code is available at https://github.com/Cynicarlos/RetinexRawMamba.

Emphasis on modeling visual context underpins the current high-performance dense prediction models, including those for change detection. However, excessive context modeling tends to cause ambiguous feature representations around object boundaries and possibly overwhelms small or thin objects. This encourages maintaining large feature maps to provide sufficient information for an objective rescue. With the goal of efficiently harvesting global context from large feature maps while getting good sensitivity to change boundaries as well as small or thin changes, we propose a change detection model that features (i) an encoder-decoder architecture with state space model-based feature refinement, and (ii) boundary-specific supervision. Our encoder-decoder is equipped with novel modulated Mamba blocks capable of preserving local correlation and then achieving local-global context mixing. To bridge the gap between local and global information during modulation, we employ a spectral transform on the local features to holistically enhance the information encoded in key frequency components. Moreover, our custom-designed boundary-specific supervision explicitly induces the revision of change boundaries. With these improvements, our Mamba-grounded change detection model can efficiently garner boundary-sensitive large feature maps applicable to various change shapes and scales. Extensive experimental results on four public change detection datasets demonstrate that our method consistently outperforms state-of-the-art competitors in terms of key evaluation metrics. Our source code is available at https://github.com/xingronaldo/BSSMamba.

Small object detection (SOD) constitutes a notable yet immensely arduous task, stemming from the restricted informative regions inherent in size-limited instances, which further sparks off heightened uncertainty beyond the capacity of current two-stage detectors. Specifically, the intrinsic ambiguity in small objects undermines the prevailing sampling paradigms and may mislead the model to devote futile effort to those unrecognizable targets, while the inconsistency of features utilized for the detection at two stages further exposes the hierarchical uncertainty. In this paper, we develop an Uncertainty learning framework for Small Object Detection, dubbed as Unc-SOD. By incorporating an auxiliary uncertainty branch to conventional Region Proposal Network (RPN), we model the indeterminacy at instance-level which later on serves as a surrogate criterion for sampling, thereby unearthing adequate candidates dynamically based on the varying degrees of uncertainty and facilitating the learning of proposal networks. In parallel, a Perception-and-Interaction strategy is devised to capture rich and discriminative representations, through optimizing the intrinsic properties from the regional features at the original pyramid and the assigned one, in which the perceptual process unfolds in a mutual paradigm. As the seminal attempt to model uncertainty in SOD task, our Unc-SOD yields state-of-the-art performance on two large-scale small object detection benchmarks, SODA-D and SODA-A, and the results on several SOD-oriented datasets including COCO, VisDrone, and Tsinghua-Tencent 100K also exhibit the promotion to baseline detector. This underscores the efficacy of our approach and its superiority over prevailing detectors when dealing with small instances.

Accurately identifying B-lines and pleural line (P-line) in lung ultrasound (LUS) videos is valuable for evaluating certain lung conditions. However, manual interpretation remains subjective and highly dependenton operator expertise. Existing deep learning methods often suffer from performance degradation due to speckle noise and motion artifacts. Moreover, the limited availability of LUS video data annotated for multiple diagnostic features such as B-lines and the P-line limits model development. Therefore, this paper introduces ILD-LUS, a new clinical LUS database designed based on interstitial lung disease (ILD) analysis by category labeling, comprising 2,149 ultrasound videos (193,410 frames). Also, we construct an external test set based on the public Covid-BLUES dataset for the evaluation of B-lines and P-line recognition in different pulmonary pathologies. Then, we propose a novel video analysis framework that integrates wavelet enhancement with temporal attention modeling. Specifically, we employ a dual-component frequency feature enhancement method using the Discrete Wavelet Transform (DWT), which effectively suppresses noise while preserving important landmarks. Subsequently, an adaptive attention module is introduced to model long-range temporal dependencies and improve dynamic feature representation across consecutive frames. Experimental results show that the proposed method achieves over 94% AUC and 82% ACC for both B-lines and P-line classification on both the ILD-LUS and Covid-BLUES datasets, outperforming existing methods. These findings demonstrate the robustness and generalizability of our approach across different pathological conditions. Overall, the proposed framework shows strong potential for supporting clinical decision-making in LUS analysis.

Automated chest X-ray report generation requires not only clinical accuracy but also transparent and interpretable diagnostic reasoning. In this work, we propose FiR-Rad, a two-stage framework that combines explicit structured reasoning with targeted fine-grained optimization. In the first stage, a supervised chain-of-thought approach guides the model to sequentially analyze and describe a comprehensive range of clinically significant thoracic abnormalities, ensuring clinically meaningful coverage. In the second stage, we introduce a segment-level reinforcement learning strategy based on Group Relative Policy Optimization (GRPO), which assigns precise rewards to each disease-specific reasoning step by evaluating the accuracy of corresponding findings in the synthesized report. This design provides direct feedback for intermediate reasoning and encourages consistency between detailed abnormality analysis and final diagnostic conclusions. Experimental results on the MIMIC-CXR and IU-Xray datasets demonstrate that our framework achieves state-of-the-art performance across clinical and linguistic metrics, with strong zero-shot generalization on IU-Xray. The proposed method significantly enhances interpretability and clinical accuracy, effectively addressing key limitations in automated radiology report generation.

Video virtual try-on aims to seamlessly dress a subject in a video with a specific garment. The primary challenge involves preserving the visual authenticity of the garment while dynamically adapting to the pose and physique of the subject. While existing methods have predominantly focused on image-based virtual try-on, extending these techniques directly to videos often results in temporal inconsistencies. Most current video virtual try-on approaches alleviate this challenge by incorporating temporal modules, yet still overlook the critical spatiotemporal pose interactions between human and garment. Effective pose interactions in videos should not only consider spatial alignment between human and garment poses in each frame but also account for the temporal dynamics of human poses throughout the entire video. With such motivation, we propose a new framework, namely Dynamic Pose Interaction Diffusion Models (DPIDM), to leverage diffusion models to delve into dynamic pose interactions for video virtual try-on. Technically, DPIDM introduces a skeleton-based pose adapter to integrate synchronized human and garment poses into the denoising network. A hierarchical attention module is then exquisitely designed to model intra-frame human-garment pose interactions and long-term human pose dynamics across frames through pose-aware spatial and temporal attention mechanisms. Moreover, DPIDM capitalizes on a temporal regularized attention loss between consecutive frames to enhance temporal consistency. Extensive experiments conducted on VITON-HD, VVT and ViViD datasets demonstrate the superiority of our DPIDM against the baseline methods. Notably, DPIDM achieves VFID score of 0.506 on VVT dataset, leading to 60.5% improvement over the state-of-the-art GPD-VVTO approach.

Audio-Visual Segmentation (AVS) faces a fundamental challenge of effectively aligning audio and visual modalities. While recent approaches leverage foundation models to address data scarcity, they often rely on single-modality knowledge or combine foundation models in an off-the-shelf manner, failing to address the cross-modal alignment challenge. In this paper, we present TAViS, a novel framework that \textbf{couples} the knowledge of multimodal foundation models (ImageBind) for cross-modal alignment and a segmentation foundation model (SAM2) for precise segmentation. However, effectively combining these models poses two key challenges: the difficulty in transferring the knowledge between SAM2 and ImageBind due to their different feature spaces, and the insufficiency of using only segmentation loss for supervision. To address these challenges, we introduce a text-bridged design with two key components: (1) a text-bridged hybrid prompting mechanism where pseudo text provides class prototype information while retaining modality-specific details from both audio and visual inputs, and (2) an alignment supervision strategy that leverages text as a bridge to align shared semantic concepts within audio-visual modalities. Our approach achieves superior performance on single-source, multi-source, semantic datasets, and excels in zero-shot settings.

The substantial computational demands of detection transformers (DETRs) hinder their deployment in resource-constrained scenarios, with the encoder consistently emerging as a critical bottleneck. A promising solution lies in reducing token redundancy within the encoder. However, existing methods perform static sparsification while ignoring the varying importance of tokens across different levels and encoder blocks for object detection, leading to suboptimal sparsification and performance degradation. In this paper, we propose **Dynamic DETR** (**Dynamic** token aggregation for **DE**tection **TR**ansformers), a novel strategy that leverages inherent importance distribution to control token density and performs multi-level token sparsification. Within each stage, we apply a proximal aggregation paradigm for low-level tokens to maintain spatial integrity, and a holistic strategy for high-level tokens to capture broader contextual information. Furthermore, we propose center-distance regularization to align the distribution of tokens throughout the sparsification process, thereby facilitating the representation consistency and effectively preserving critical object-specific patterns. Extensive experiments on canonical DETR models demonstrate that Dynamic DETR is broadly applicable across various models and consistently outperforms existing token sparsification methods.

Novel-view synthesis approaches play a critical role in vast scene reconstruction. However, these methods rely heavily on dense image inputs and prolonged training times, making them unsuitable where computational resources are limited. Additionally, few-shot methods often struggle with poor reconstruction quality in vast environments. This paper presents DGTR, a novel distributed framework for efficient Gaussian reconstruction for sparse-view vast scenes. Our approach divides the scene into regions, processed independently by drones with sparse image inputs. Using a feed-forward Gaussian model, we predict high-quality Gaussian primitives, followed by a global alignment algorithm to ensure geometric consistency. Depth priors is incorporated to further enhance training, while a distillation-based model aggregation mechanism enables efficient reconstruction. Our method achieves high-quality large-scale scene reconstruction and novel-view synthesis in significantly reduced training times, outperforming existing approaches in both speed and scalability. We demonstrate the effectiveness of our framework on vast aerial scenes, achieving high-quality results within minutes. Code will released on our project page https://3d-aigc.github.io/DGTR.

Efficient trackers achieve faster runtime by reducing computational complexity and model parameters. However, this efficiency often compromises the expense of weakened feature representation capacity, thus limiting their ability to accurately capture target states using single-layer features. To overcome this limitation, we propose Multi-State Tracker (MST), which utilizes highly lightweight state-specific enhancement (SSE) to perform specialized enhancement on multi-state features produced by multi-state generation (MSG) and aggregates them in an interactive and adaptive manner using cross-state interaction (CSI). This design greatly enhances feature representation while incurring minimal computational overhead, leading to improved tracking robustness in complex environments. Specifically, the MSG generates multiple state representations at multiple stages during feature extraction, while SSE refines them to highlight target-specific features. The CSI module facilitates information exchange between these states and ensures the integration of complementary features. Notably, the introduced SSE and CSI modules adopt a highly lightweight hidden state adaptation-based state space duality (HSA-SSD) design, incurring only 0.1 GFLOPs in computation and 0.66 M in parameters. Experimental results demonstrate that MST outperforms all previous efficient trackers across multiple datasets, significantly improving tracking accuracy and robustness. In particular, it shows excellent runtime performance, with an AO score improvement of 4.5% over the previous SOTA efficient tracker HCAT on the GOT-10K dataset. The code is available at https://github.com/wsumel/MST.

The Zero-shot Vision-and-Language Navigation in Continuous Environments (VLN-CE) task requires agents to navigate previously unseen 3D environments using natural language instructions, without any scene-specific training. A critical challenge in this setting lies in ensuring agents’ actions align with both spatial structure and task intent over long-horizon execution. Existing methods often fail to achieve robust navigation due to a lack of structured decision-making and insufficient integration of feedback from previous actions. To address these challenges, we propose STRIDER (Instruction-Aligned Structural Decision Space Optimization), a novel framework that systematically optimizes the agent’s decision space by integrating spatial layout priors and dynamic task feedback. Our approach introduces two key innovations: 1) a Structured Waypoint Generator that constrains the action space through spatial structure, and 2) a Task-Alignment Regulator that adjusts behavior based on task progress, ensuring semantic alignment throughout navigation. Extensive experiments on the R2R-CE and RxR-CE benchmarks demonstrate that STRIDER significantly outperforms strong SOTA across key metrics; in particular, it improves Success Rate (SR) from 29\% to 35\%, a relative gain of 20.7\%. Such results highlight the importance of spatially constrained decision-making and feedback-guided execution in improving navigation fidelity for zero-shot VLN-CE.

MOTIVATION: Alzheimer's disease (AD) typically progresses gradually for ages rather than suddenly. Thus, staging AD progression in different phases could aid in accurate diagnosis and treatment. In addition, identifying genetic variations that influence AD is critical to understanding the pathogenesis. However, staging the disease progression and identifying genetic variations is usually handled separately. RESULTS: To address this limitation, we propose a novel sparse multi-stage multi-task mixed-effects collaborative longitudinal regression method (MSColoR). Our method jointly models long disease progression as a multi-stage procedure and identifies genetic risk factors underpinning this complex trajectory. Specifically, MSColoR models multi-stage disease progression using longitudinal neuroimaging-derived phenotypes and associates the fitted disease trajectories with genetic variations at each stage. Furthermore, we collaboratively leverage summary statistics from large genome-wide association studies to improve the powers. Finally, an efficient optimization algorithm is introduced to solve MSColoR. We evaluate our method using both synthetic and real longitudinal neuroimaging and genetic data. Both results demonstrate that MSColoR can reduce modeling errors while identifying more accurate and significant genetic variations compared to other longitudinal methods. Consequently, MSColoR holds great potential as a computational technique for longitudinal brain imaging genetics and AD studies. AVAILABILITY AND IMPLEMENTATION: The code is publicly available at https://github.com/dulei323/MSColoR.

Most existing CoSOD models focus solely on extracting co-saliency cues while neglecting explicit exploration of background regions, potentially leading to difficulties in handling interference from complex background areas. To address this, this paper proposes a Discriminative co-saliency and background Mining Transformer framework (DMT) to explicitly mine both co-saliency and background information and effectively model their discriminability. DMT first learns two types of tokens by disjointly extracting co-saliency and background information from segmentation features, then performs discriminability within the segmentation features guided by these well-learned tokens. In the first phase, we propose economic multi-grained correlation modules for efficient detection information extraction, including Region-to-Region (R2R), Contrast-induced Pixel-to-Token (CtP2T), and Co-saliency Token-to-Token (CoT2T) correlation modules. In the subsequent phase, we introduce Token-Guided Feature Refinement (TGFR) modules to enhance discriminability within the segmentation features. To further enhance the discriminative modeling and practicality of DMT, we first upgrade the original TGFR's intra-image modeling approach to an intra-group one, thus proposing Group TGFR (G-TGFR), which is more suitable for the co-saliency task. Subsequently, we designed a Noise Propagation Suppression (NPS) mechanism to apply our model to a more practical open-world scenario, ultimately presenting our extended version, i.e. DMT+O. Extensive experimental results on both conventional CoSOD and open-world CoSOD benchmark datasets demonstrate the effectiveness of our proposed model.

Few-shot segmentation (FSS) aims to accurately segment target objects in a query image using only a limited number of annotated support images. Existing approaches typically follow a paradigm that directly leverages category information from the support set to identify target objects in the query. However, these methods often ignore the category information gap between query and support images, leading to suboptimal performance when faced with images containing objects exhibiting significant intra-class diversity. To address this issue, we propose a novel framework that introduces intermediate prototypes to capture both deterministic information from the support images and adaptive knowledge from the query at multiple scales. Our framework, named the K-shot Multi-scale Intermediate Prototype Mining Transformer (KMIPMT), is based on the Transformer architecture and learns intermediate prototypes in an iterative manner, where each KMIPMT layer propagates category information from both K-shot support features and multi-scale query features to intermediate prototypes. This information is then utilized to activate the query feature map. Through repeated iterations, both intermediate prototypes and the query feature are progressively enhanced, and the final refined query feature is used for generating precise segmentation predictions. Despite its simplicity, our method achieves remarkable performance gains on standard benchmarks, including PASCAL-$5^{i}$5i, COCO-$20^{i}$20i, and FSS-1000, setting new state-of-the-art results. Furthermore, we explore several practical and challenging extensions of our method, including 3D point cloud FSS, zero-shot segmentation, weak-label FSS, and cross-domain FSS. These extensions showcase the versatility and effectiveness of our proposed KMIPMT framework across different domains and scenarios.

In this paper, we explain the mechanism of bilinear pooling as a module of hard sample generation, and find that bilinear pooling significantly expands variances of the first-order vectors when it produces discriminative bilinear features. In conjunction with the extremely high dimensionality of the obtained bilinear features, those variances lead to overfitting in subsequent learning models. To solve this issue, we construct a bi-level optimization problem, where the high-level problem is the supervised classification loss, and the low-level problem is the principal component analysis (PCA). Then, we find that PCA on bilinear features is equivalent to spectral clustering, which allows us to mathematically prove that the first $\log _{2}(C)$log2(C) principal components can support the discriminant information of $C$C classes. By removing the rest principal components, the dimensionality and variances are simultaneously reduced. To the best of our knowledge, this is the first work providing a lower bound for dimension reduction for bilinear pooling. However, the PCA projection matrix $\mathbf{L}$L is prone to overfitting due to having many parameters. To address this issue, we propose a rank-$k$k general bilinear projection (RK-GBP) that decomposes $\mathbf{L}$L into two small matrices $\mathbf{U}$U and $\mathbf{V}$V, whose learnable parameters are smaller. Different from traditional bilinear projections used in factorized bilinear pooling (FBiP), our RK-GBP can preserve the orthogonality of columns in $\mathbf{L}$L by constraining the orthogonality of columns in $\mathbf{U}$U and $\mathbf{V}$V. For computational efficiency, we relax the PCA in the low-level task into a dictionary learning problem, obtaining the rank-$k$k orthogonal factorization bilinear pooling (RK-OFBP). The RK-OFBP can be considered as a general form of current factorization bilinear pooling methods (e.g., Hadamard product-based ones). Finally, we evaluate our approach on fine-grained images and large-scale datasets, demonstrating that our proposed method not only produces extremely low-dimensional features but also outperforms other methods in classification tasks. For example, our RK-OFBP can employ 32-dimensional vectors to achieve comparable results to B-CNN (Lin, 2015) (dimension: 512*512) for the 200-class classification task.

On account of the extreme settings, stealing the black-box model without its training data is difficult in practice. On this topic, along the lines of data diversity, this paper substantially makes the following improvements based on our conference version (dubbed STDatav1, short for Surrogate Training Data). First, to mitigate the undesirable impacts of the potential mode collapse while training the generator, we propose the joint-data optimization scheme, which utilizes both the synthesized data and the proxy data to optimize the surrogate model. Second, we propose the self-conditional data synthesis framework, an interesting effort that builds the pseudo-class mapping framework via grouping class information extraction to hold the class-specific constraints while holding the diversity. Within this new framework, we inherit and integrate the class-specific constraints of STDatav1 and design a dual cross-entropy loss to fit this new framework. Finally, to facilitate comprehensive evaluations, we perform experiments on four commonly adopted datasets, and a total of eight kinds of models are employed. These assessments witness the considerable performance gains compared to our early work and demonstrate the competitive ability and promising potential of our approach.

In recent times, following the paradigm of DETR (DEtection TRansformer), query-based end-to-end instance segmentation (QEIS) methods have exhibited superior performance compared to CNN-based models, particularly when trained on large-scale datasets. Nevertheless, the effectiveness of these QEIS methods diminishes significantly when confronted with limited training data. This limitation arises from their reliance on substantial data volumes to effectively train the pivotal queries/kernels that are essential for acquiring localization and shape priors. To address this problem, we propose a novel method for unsupervised pre-training in low-data regimes. Inspired by the recently successful prompting technique, we introduce a new method, Unsupervised Pre-training with Language-Vision Prompts (UPLVP), which improves QEIS models' instance segmentation by bringing language-vision prompts to queries/kernels. Our method consists of three parts: (1) Masks Proposal: Utilizes language-vision models to generate pseudo masks based on unlabeled images. (2) Prompt-Kernel Matching: Converts pseudo masks into prompts and injects the best-matched localization and shape features to their corresponding kernels. (3) Kernel Supervision: Formulates supervision for pre-training at the kernel level to ensure robust learning. With the help of our pre-training method, QEIS models can converge faster and perform better than CNN-based models in low-data regimes. Experimental evaluations conducted on MS COCO, Cityscapes, and CTW1500 datasets indicate that the QEIS models' performance can be significantly improved when pre-trained with our method.

Eye-tracking is a reliable method for quantifying visual information processing and holds significant potential for group recognition, such as identifying autism spectrum disorder (ASD). However, eye-tracking research typically faces the heterogeneity of stimuli and is time-consuming due to the large number of observed stimuli. To address these issues, we first mathematically define the stimulus selection problem and introduce the concept of stimulus discrimination ability to reduce the computational complexity of the solution. Then, we construct a scanpath-based recognition model to mine the stimulus discrimination ability. Specifically, we propose cross-subject entropy and cross-subject divergence scores for quantitatively evaluating stimulus discrimination ability, effectively capturing differences in intra-group collective trends and inter-subject consistency within a group. Furthermore, we propose an iterative learning mechanism that employs stimulus-wise attention to focus on discriminative stimuli for discrimination purification. In the experiment, we construct an ASD eye-tracking dataset with diverse stimulus types and conduct extensive tests on three representative models to validate our approach. Remarkably, our method demonstrates superior performance using only 10 selected stimuli compared to models utilizing 220 stimuli. Additionally, we perform experiments on another eye-tracking task, gender prediction, to further validate our method. We believe that our approach is both simple and flexible for integration into existing models, promoting large-scale ASD screening and extending to other eye-tracking research domains.

Recognizing autism spectrum disorder (ASD) has faced great challenges due to insufficient professional clinicians and complex procedures. Automated data-driven ASD recognition models can reduce the subjectivity and physician dependency of traditional evaluation methods. Facial data, which can encode important perceptual and social behaviors, have emerged in ASD research to explore novel biomarkers for screening, diagnosing, and treating ASD. However, existing research mainly focuses on extracting low-level hand-crafted facial features for analysis and classification. Determining how to learn discriminative deep representations from dynamic facial data for computational model construction remains an unresolved challenge. In this study, we propose an ASD recognition model based on facial videos to fill the lack of temporal correlation learning of facial features. First, we utilize a vision transformer to extract frame-based global facial features. Then, we use a Longformer to establish the correlation of facial features over time. In the experiment, we recruited 146 subjects between 2 and 8 years of age to record their facial videos under a computer-based eye-tracking experiment and 76 subjects to conduct a smartphone-based experiment. Quantitative comparisons have shown the effectiveness and reliability of the proposed model. Furthermore, we have confirmed the correlation between facial and eye-tracking modalities in visual attention.

The disperse structure distributions (discreteness) and variant scattering characteristics (variability) of SAR airplane targets lead to special challenges of object detection and recognition. The current deep learning-based detectors encounter challenges in distinguishing fine-grained SAR airplanes against complex backgrounds. To address it, we propose a novel physics-guided detector (PGD) learning paradigm for SAR airplanes that comprehensively investigate their discreteness and variability to improve the detection performance. It is a general learning paradigm that can be extended to different existing deep learning-based detectors with ”backbone-neck-head” architectures. The main contributions of PGD include the physics-guided self-supervised learning, feature enhancement, and instance perception, denoted as PGSSL, PGFE, and PGIP, respectively. PGSSL aims to construct a self-supervised learning task based on a wide range of SAR airplane targets that encodes the prior knowledge of various discrete structure distributions into the embedded space. Then, PGFE enhances the multi-scale feature representation of a detector, guided by the physics-aware information learned from PGSSL. PGIP is constructed at the detection head to learn the refined and dominant scattering point of each SAR airplane instance, thus alleviating the interference from the complex background. We propose two implementations, denoted as PGD and PGD-Lite, and apply them to various existing detectors with different backbones and detection heads. The experiments demonstrate the flexibility and effectiveness of the proposed PGD, which can improve existing detectors on SAR airplane detection with fine-grained classification task (an improvement of 3.1% mAP most), and achieve the state-of-the-art performance (90.7% mAP) on SAR-AIRcraft-1.0 dataset. The project is open-source at https://github.com/XAI4SAR/PGD.

Few-shot semantic segmentation (FSS) aims to segment the target object under the condition of a few annotated samples. However, current studies on FSS primarily concentrate on extracting information related to the object, resulting in inadequate identification of ambiguous regions, particularly in non-target areas, including the background (BG) and Distracting Objects (DOs). Intuitively, to alleviate this problem, we propose a novel framework, namely NTRENet++, to explicitly mine and eliminate BG and DO regions in the query. First, we introduce a BG Mining Module (BGMM) to extract BG information and generate a comprehensive BG prototype from all images. For this purpose, a BG mining loss is formulated to supervise the learning of BGMM, utilizing only the known target object segmentation ground truth. Subsequently, based on this BG prototype, we employ a BG Eliminating Module to filter out the BG information from the query and obtain a BG-free result. Following this, the target information is utilized in the target matching module to generate the initial segmentation result. Finally, a DO Eliminating Module is proposed to further mine and eliminate DO regions, based on which we can obtain a BG and DO-free target object segmentation result. Moreover, we present a prototypical-pixel contrastive learning algorithm to enhance the model’s capability to differentiate the target object from DOs. Extensive experiments conducted on both PASCAL-5i and COCO-20i datasets demonstrate the effectiveness of our approach despite its simplicity. Additionally, we extend our method to the few-shot video object segmentation task and achieve improved performance on a baseline model, demonstrating its generalization ability. Code is available athttps://github.com/LIUYUANWEI98/NTRENet++.

Fine-grained object detection (FGOD) in remote sensing images is an emerging and challenging task in the field of image intelligent interpretation. It aims to localize objects while classifying them into different fine-grained categories. Modern FGOD methods are mainly derived from well-developed detectors and have made compelling progress. Despite this, these methods struggle to perform well in classifying objects at the subordinate level due to the limitations of their representation manners. In this paper, we propose a network capable of learning discriminative representation (DR) for fine-grained object detection in remote sensing images, named DRNet. First, a fine-grained branch that works in parallel with other task branches is introduced, where objects’ features are re-encoded with dual refinement to generate discriminative representation, enabling accurate fine-grained classification. Second, we design a confusion-minimized loss that automatically scales loss contributions according to the separability of samples to train the fine-grained branch, further boosting discriminative ability of the representation and better addressing hard-to-distinguish objects. Moreover, we devise an interaction verification strategy that empowers the network to fully utilize the results of fine-grained classification and coarse classification for achieving robust inference. On large-scale FAIR1M-1.0 and FAIR1M-2.0 datasets, our DRNet with ResNet50 and$1\times $training schedule obtains 40.87% mAP and 47.04% mAP, respectively, establishing new state-of-the-arts for fine-grained object detection in remote sensing images. The source code is available athttps://github.com//54wb//DRNet.

In object detection, particularly within remote sensing images, the quality of selected samples is crucial for the accuracy and robustness of detection models. However, current sampling strategies demonstrate inherent limitations. They empirically define positive sample sets using fixed thresholds or preset areas, ignoring the actual shapes of the objects and failing to distinguish the intrinsic value of each sample point. To address these critical issues, this article proposes a novel centric probability-based sample selection approach that includes centering probability mapping (CPM), Expectation-Maximization-based boundary optimization (EBO), and probabilistic random sampling (PRS) technologies. Specifically, the CPM is constructed to assign various confidence levels for all sample points based on their proximity to the center of bounding box, effectively discerning the value of individual samples. Then, the EBO is utilized to dynamically optimize the boundaries for positive and negative samples based on the EM algorithm, thus avoiding the sample imbalance problem associated with empirical thresholds. Finally, the PRS strategy is proposed to select training samples from the sample space constructed by CPM and EBO in a manner of random probability sampling, which could improve the diversity of samples while guaranteeing their quality. Experimental validation on three remote sensing image datasets, including DOTA-v1.0, DOTA-v2.0, and DIOR-R, demonstrates that our method achieves robust performance improvements over baseline and significantly surpasses the advanced sample selection methods. The source code will be available athttps://github.com/yanqingyao1994/CPSS.

Detecting small, oriented objects in remote sensing images remains a bottleneck for prevailing detection paradigms. The discriminative cues essential for detecting small instances are often inaccessible owing to the restrained spatial extent and poor visual responses, which further compromises the model and necessitates reliance on low-level patterns for identification and localization, exacerbating vulnerability to structural distortions and intra-class confusion especially in complex scenarios. To address these desiderata, we devise a Semantic Differentiation (SemDiff) framework for oriented small object detection in remote sensing images. Starting with randomly initialized category-specific units, we deliver a differentiation pipeline where distinctive features steer the evolution of these embeddings via a tailored differentiation loss. Afterwards, these class-aligned vectors function as dynamic kernels, infusing hierarchical representations with semantic understanding. Moreover, an improved centerness metric that is more accommodating to size-constrained instances is introduced. Building upon this, we design an instance-level recalibration mechanism to regulate the training process, thereby ensuring adequate optimization even for exceptionally small instances. By integrating semantic in an explicit fashion, our SemDiff efficiently facilitates the discriminative capabilities of hierarchical features, thereby revitalizing foreground responses and alleviating semantic-level ambiguity. On the challenging small object detection benchmarks SODA-A and Tiny-DOTA, our approach outstrips prevailing single-stage paradigms by a substantial margin, and achieves competitive performance to its two-stage counterparts, but with an edge of speed. Codes will be available athttps://github.com/shaunyuan22/SemDiff.

Few-shot object detection (FSOD) in remote sensing images is a marginally explored but highly challenging task that focuses on identifying unseen classes of objects with a limited number of annotations. Current FSOD approaches often fail to accurately localize the foreground and misalign targets with various orientations, resulting in poor detection performance. For this purpose, we develop a fresh and powerful meta-learning framework based on the idea of imprinting, which leverages tailored support information to model the regional correlation between query and support objects in different stages. Specifically, a global-integrated scheme is first proposed to guide the generation of high-quality proposals by increasing the activation of foreground features and integrating global support information. Considering the orientation discrepancy of objects in query and support sets, we introduce a drift-rectified technique to achieve adaptive alignment by implicitly capturing the positional correspondence between the instances in two sets. In stark contrast to conventional FSOD approaches, our method can extract key clues and establish directional relationships between objects from different training sets, leading to better generalization capability. Extensive experiments on two standard benchmarks (DIOR and NWPU VHR-10.V2) manifest the effectiveness, and our proposed method exhibits superior performance to other competitors with similar motivation. The source code is available athttps://github.com/Ybowei/GIDR

Within remote sensing images, complex atmospheric environments commonly bring about distinct variations in imaging visibility and ambient occlusions, significantly transforming the appearance of objects. Nevertheless, modern detectors generally struggle to maintain promising accuracy when encountering realistic scenarios. Devoting to alleviating the issues, we develop a noise incentive robust network (NIRNet) for remote sensing object detection under cloud corruption without relying on hazy images for training. The proposed NIRNet preserves discriminative representations and calibrates them using an incentive mechanism. Firstly, we design a noise perception module (NPM) to deal with diverse cloud corruption types, which generates point-wise calibration weights dependent on the perceived discrepancy between objects and environmental noise. Secondly, aiming to detect difficult-to-discern objects thoroughly, a dual-path incentive calibration (DPIC) strategy is proposed to combine intensity and stability features weighted by NPM. Profiting from its universal design, the DPIC could be treated as a plug-and-play module for existing detectors, enhancing robustness against adverse weather. To evaluate the reliability of aerial detectors under intricate cloud corruptions, we present an elaborate Hazy-DIOR dataset, which contains numerous images with different cloud conditions and severity levels. Finally, extensive experiments on the Hazy-DIOR and DOTA-Cloud datasets simultaneously demonstrate the robustness of NIRNet, which especially achieves state-of-the-art accuracy and gets 2.16% mAP and 2.56% rPC improvements on the Hazy-DIOR compared to solid Oriented R-CNN detector. The code is available at https://github.com/zhangpeng2001/nirnet.

Synthetic aperture radar (SAR) imaging provides a distinct advantage in scene understanding due to its capability for all-weather data acquisition. However, in comparison to easily annotated optical remote sensing images, the lower imaging quality of SAR images presents significant challenges in obtaining manually annotated training data, which poses substantial issues for SAR image analysis. In this paper, we employ the domain adaptation (DA) that leverages labeled optical images to better understand unlabeled SAR images. Global feature alignment as a method for DA has demonstrated effectiveness in transferring knowledge, yet it faces challenges in cross-modality adaptation from optical remote sensing to SAR images due to their differing imaging mechanisms. With distinct visual features between optical and SAR images, the semantic dependency is difficult to construct, which results in low-quality pseudo-label assignment for SAR images. To address the above issue, we propose a semantic graph learning framework to comprehensively align the global features of optical remote sensing and SAR images by modeling the cross-modality semantics and generating high-quality pseudo-labels. It can be applied for SAR scene classification and object detection when only optical remote sensing images are labeled. Specifically, a cross-modality semantic graph alignment (CSGA) module is constructed to model and align the second-order semantic dependencies by aggregating cross-modality visual semantic information. Then, an uncertainty-based robust pseudo-label generation (URPG) module is designed to generate pseudo-labels for effective semantic alignment and self-training by modeling the uncertainty of pseudo-labels for each SAR image. Comprehensive experiments show that our proposed method outperforms the state-of-the-art methods on scene classification (NWPU-RESISC45→WHU-SAR6, MLRSNet→NWPU-SAR6, MLRSNet→NWPU-SAR6, and NWPU-RESISC45→NWPU-SAR6) and object detection (MASATI-ship→SSDD, MVSRD→SARDet-vehicle, and DIOR-airplane→SAR-airplane) tasks. The code and datasets are publicly accessible at https://github.com/XZhang878/SGLF.

Synthetic aperture radar (SAR) image simulation has attracted much attention due to its great potential to supplement the scarce training data for deep learning algorithms. Consequently, evaluating the quality of the simulated SAR image is crucial for practical applications. The current literature primarily uses image quality assessment (IQA) techniques for evaluation that rely on human observers' perceptions. However, because of the unique imaging mechanism of SAR, these techniques may produce evaluation results that are not entirely valid. The distribution inconsistency between real and simulated data is the main obstacle that influences the utility of simulated SAR images. To this end, we propose a novel trustworthy utility evaluation framework with a counterfactual explanation for simulated SAR images for the first time, denoted as X-Fake. It unifies a probabilistic evaluator and a causal explainer to achieve a trustworthy utility assessment. We construct the evaluator using a probabilistic Bayesian deep model to learn the posterior distribution, conditioned on real data. Quantitatively, the predicted uncertainty of simulated data can reflect the distribution discrepancy. We build the causal explainer with an introspective variational auto-encoder (IntroVAE) to generate high-resolution counterfactuals. The latent code of IntroVAE is finally optimized with evaluation indicators and prior information to generate the counterfactual explanation, thus revealing the inauthentic details of simulated data explicitly. The proposed framework is validated on four simulated SAR image datasets obtained from electromagnetic models and generative artificial intelligence approaches. The results demonstrate the proposed X-Fake framework outperforms other IQA methods in terms of utility. Furthermore, the results illustrate that the generated counterfactual explanations are trustworthy, and can further improve the data utility in applications.

Weakly supervised semantic segmentation (WSSS) is a challenging yet important research field in vision community. In WSSS, the key problem is to generate high-quality pseudo segmentation masks (PSMs). Existing approaches mainly depend on the discriminative object part to generate PSMs, which would inevitably miss object parts or involve surrounding image background, as the learning process is unaware of the full object structure. In fact, both the discriminative object part and the full object structure are critical for deriving of high-quality PSMs. To fully explore these two information cues, we build a novel end-to-end learning framework, alternate self-dual teaching (ASDT), based on a dual-teacher single-student network architecture. The information interaction among different network branches is formulated in the form of knowledge distillation (KD). Unlike the conventional KD, the knowledge of the two teacher models would inevitably be noisy under weak supervision. Inspired by the Pulse Width (PW) modulation, we introduce a PW wave-like selection signal to alleviate the influence of the imperfect knowledge from either teacher model on the KD process. Comprehensive experiments on the PASCAL VOC 2012 and COCO-Stuff 10K demonstrate the effectiveness of the proposed ASDT framework, and new state-of-the-art results are achieved.

Foundational models have significant potential to advance brain function research, particularly in understanding the dynamics of brain states. However, most existing models process brain signals within fixed time windows, restricting their ability to capture the full temporal complexity of brain activity. In this study, we propose BrainSN (Brain States Network), a novel fMRI foundational model designed to represent continuous brain state information and support diverse downstream tasks. First, leveraging a transformer-based architecture, BrainSN reconstructs input brain states across multiple time scales and predicts future brain activity, effectively capturing both short-term and long-term dependencies. Second, through multiple embeddings and a channel gating module, the model integrates brain state information and applies an attention mechanism to extract critical features. Additionally, we train BrainSN on 1,256 hours of resting-state and naturalistic stimulus fMRI data, enabling it to learn large-scale brain dynamics without relying on task-based paradigms. Without fine-tuning, BrainSN achieves 75.23% and 75.82% accuracy in autism and attention disorder diagnosis tasks, respectively, matching the performance of leading models pretrained on disease-specific data. After fine-tuning, it surpasses these models. In mental state decoding, BrainSN attains 95.31% accuracy without fine-tuning, outperforming the best models trained on large-scale task-based fMRI data. Furthermore, by analyzing BrainSN's embeddings in relation to movie stimuli, we demonstrate that the model effectively captures the semantic content of movie scenes embedded in fMRI signals and is highly sensitive to sequence. These results highlight BrainSN's ability to model brain state dynamics and underscore its potential advantages for clinical diagnosis, treatment evaluation, and cognitive neuroscience research.

Glomerulus detection is a critical component of renal histopathology assessment, essential for diagnosing glomerulonephritis. To mitigate the increasing workload on pathologists, AI-assisted diagnostic methods based on high-resolution digital pathology whole slide images have been developed. However, these current AI-assisted approaches are limited to high-resolution whole slide images, necessitating expensive digital scanner equipment, high image storage costs, and significant computational complexity. To address this limitation, this paper pioneers a method for facilitating glomerulus detection in low-resolution human kidney pathology images. Specifically, we propose a novel multi-step hybrid knowledge distillation method. Our method distills both the global features and the semantic information through a hybrid knowledge distillation strategy that integrates offline and online knowledge distillation, where the information from high-resolution pathological images is successively transferred to student model from the global features in the shallow network layers to the semantic information of the back-end through a multi-step training strategy. Experimental results on two datasets show that the proposed method achieves effective detection outcomes for low-resolution kidney pathology images. Compared to other state-of-the-art detection techniques, our method achieves an ${AP}_{0.5:0.95}$ improvement of 23.1% on the private LN dataset and 15.9% on the public HUBMAP dataset.

Predicting individual-level non-neuroimaging phenotypes (e.g., fluid intelligence) using brain imaging data is a fundamental goal of neuroscience. Recent research has focused on utilizing high-cost functional magnetic resonance imaging (fMRI) to predict phenotypes seen during training. However, these methods 1) only consider predicting seen phenotypes, failing to achieve zero-shot inference for unseen phenotypes; 2) overlook the knowledge transfer from fMRI to structural MRI (sMRI), missing out on utilizing cost-effective sMRI for accurate predictions. To address these challenges, we propose a Foundational Model for Any Phenotype Prediction via fMRI to sMRI knowledge transfer (FM-APP), consisting of a Phenotypes Text Memory Bank (PTMB) module, Any Phenotype Prediction (APP) module, and fMRI to sMRI Knowledge Transfer (F2SKT) module. Our proposed FM-APP adapts to downstream tasks by generating regressor parameters instead of fine-tuning the model itself. Specifically, to retain important clues from seen phenotype descriptions, PTMB utilizes the BiomedCLIP model to store semantic features of seen phenotypes. To achieve any phenotype prediction, the APP introduces a regressor synthesizer for zero-shot inference. Additionally, to improve sMRI prediction accuracy while preserving its cost advantage, the F2SKT uses the PTMB to construct phenotype active maps, guiding adaptive knowledge transfer from fMRI to sMRI. Experiments on the Human Connectome Project (HCP) and HCP Aging datasets demonstrate our approach outperforms state-of-the-art methods, showcasing strong zero-shot inference capabilities and providing a novel framework for analyzing brain structure and phenotypes. Our code: https://github.com/ZhibinHe/FM-APP.

Nuclei instance segmentation and classification are a fundamental and challenging task in whole slide Imaging (WSI) analysis. Most dense nuclei prediction studies rely heavily on crowd labelled data on high-resolution digital images, leading to a time-consuming and expertise-required paradigm. Recently, Vision-Language Models (VLMs) have been intensively investigated, which learn rich cross-modal correlation from large-scale image-text pairs without tedious annotations. Inspired by this, we build a novel framework, called PromptNu, aiming at infusing abundant nuclei knowledge into the training of the nuclei instance recognition model through vision-language contrastive learning and prompt engineering techniques. Specifically, our approach starts with the creation of multifaceted prompts that integrate comprehensive nuclear knowledge, including visual insights from the GPT-4V model, statistical analyses, and expert insights from the pathology field. Then, we propose a novel prompting methodology that consists of two pivotal vision-language contrastive learning components: the Prompting Nuclei Representation Learning (PNuRL) and the Prompting Nuclei Dense Prediction (PNuDP), which adeptly integrates the expertise embedded in pre-trained VLMs and multifaceted prompts into the feature extraction and prediction process, respectively. Comprehensive experiments on six datasets with extensive WSI scenarios demonstrate the effectiveness of our method for both nuclei instance segmentation and classification tasks. The code is available at https://github.com/NucleiDet/PromptNu.

As two intimate reciprocal tasks, scene-aware human motion synthesis and analysis require a joint understanding between multiple modalities, including 3D body motions, 3D scenes, and textual descriptions. In this paper, we integrate these two paired processes into a Co-Evolving Synthesis-Analysis (CESA) pipeline and mutually benefit their learning. Specifically, scene aware text-to-human synthesis generates diverse indoor motion samples from the same textual description to enrich human scene interaction intra-class diversity, thus significantly benefiting training a robust human motion analysis system. Reciprocally, human motion analysis would enforce semantic scrutiny on each synthesized motion sample to ensure its semantic consistency with the given textual description, thus improving realistic motion synthesis. Considering that real-world indoor human motions are goal-oriented and path-guided, we propose a cascaded generation strategy that factorizes text-driven scene-specific human motion generation into three stages: goal inferring, path planning, and pose synthesizing. Coupling CESA with this powerful cascaded motion synthesis model, we jointly improve realistic human motion synthesis and robust human motion analysis in 3D scenes.

Medical report generation refers to the automatic creation of accurate and coherent diagnostic reports for medical images. This task can alleviate the workload of radiologists, enhance the efficiency of disease diagnosis, and therefore holds significant value and challenges. Considering the feature differences between different modalities, existing methods primarily focus on facilitating medical report generation through cross-modal alignment of images and texts. However, since medical images are very similar to each other, it is difficult to tag obvious objects, making most methods limited to coarse-grained image-text global alignment. In this paper, we propose a medical report generation model based on adaptive topic learning and fine-grained cross-modal alignment, which aligns images and texts from medical topic perspective and token perspective. From the medical topic perspective, a global-local contrastive loss is introduced to adaptively learn efficient medical topic features, and medical topics are utilized to map images and texts to the same semantic space for fine-grained alignment. From the token perspective, a token prediction module is designed to enable the model to focus on important local information by predicting the key tokens contained in the report. Experimental results on the two public datasets (i.e. IU-Xray and MIMIC-CXR) demonstrate that our proposed model outperforms state-of-the-art baselines.

Large language models (LLMs) such as ChatGPT have recently demonstrated significant potential in mathematical abilities, providing a valuable reasoning paradigm consistent with human natural language. However, LLMs currently have difficulty in bridging perception, language understanding, and reasoning (PLR) capabilities due to incompatibility of the underlying information flow among them, making their reasoning ability not fully elicited and challenging to accomplish complicated reasoning tasks autonomously. To resolve the above problem, a novel method called ChatABL is proposed by integrating LLMs into an abductive learning (ABL) framework, capable of unifying the three abilities effectively in a more user-friendly and understandable manner. Initially, the proposed method uses LLMs to correct the incomplete logical facts for optimizing the perception module, by summarizing and reorganizing domain knowledge represented in natural language format. Then, the perception module also provides necessary logical reasoning materials for feeding LLMs. Finally, these parts are integrated into a dynamic closed-loop system by introducing the feedback form and automatic learning strategies to mutually promote their performance. As a testbed, the variable-length handwritten equation decipherment (HED), an abstract expression of the Mayan calendar decoding, is used to demonstrate that ChatABL has reasoning ability beyond most existing state-of-the-art methods, which has been well-supported by comparative studies. To the best of authors' knowledge, the proposed ChatABL is the first attempt to explore a possible and novel avenue to approaching human-level cognitive ability via natural language interaction by means of ChatGPT.

Saccadic scanpath, a data representation of human visual behavior, has received broad interest in multiple domains. Scanpath is a complex eye-tracking data modality that includes the sequences of fixation positions and fixation duration, coupled with image information. However, previous methods usually face the spatial misalignment problem of fixation features and loss of critical temporal data (including temporal correlation and fixation duration). In this study, we propose a Transformer-based scanpath model, SpFormer, to alleviate these problems. First, we propose a fixation-centric paradigm to extract the aligned spatial fixation features and tokenize the scanpaths. Then, according to the visual working memory mechanism, we design a local meta attention to reduce the semantic redundancy of fixations and guide the model to focus on the meta scanpath. Finally, we progressively integrate the duration information and fuse it with the fixation features to solve the problem of ambiguous location with the Transformer block increasing. We conduct extensive experiments on four databases under three tasks. The SpFormer establishes new state-of-the-art results in distinct settings, verifying its flexibility and versatility in practical applications. The code can be obtained from https://github.com/wenqizhong/SpFormer.

Radiological Report Summarization (RRS) involves automated summarization of key impressions derived from identified findings, intending to alleviate the workload and stress experienced by radiologists. Many existing RRS methods predominantly concentrate on summarizing findings, neglecting crucial clinical context, such as the patient’s previous medical examinations. This context, which is a focal point for radiologists, plays a critical role in producing comprehensive and accurate impressions. This paper endeavors to emulate the workflows of radiologists by incorporating the patient’s clinical context alongside current findings. To achieve this, we reconceptualize RRS as a conversational question-answering task, generating temporal radiological conversations. These conversations are subsequently employed to fine-tune a large chat model. The resulting radiology chatbot, RadChat, demonstrates superior performance in RRS task, showcasing the potential of integrating clinical context for more accurate impressions. Experimental results conducted on the MIMIC-CXR dataset validate the superiority of RadChat in comparison to state-of-the-art baselines.

Brain imaging genetics offers insights into the genetic basis of brain structure and function by exploring the relationships between genetic variations and neuroimaging features, with canonical correlation association learning as a vital and effective tool. However, imaging large cohorts affected by specific brain diseases entails significant costs. To tackle this challenge, we introduced a novel bi-multivariate sparse canonical correlation association method based on summary statistics from large GWAS (S-SCCA). S-SCCA leverages effect sizes obtained from these datasets to identify genetic variants associated with complex traits, including those influenced by pleiotropy, while simultaneously identifying imaging factors related to the disease under study. Moreover, we have implemented a rapid optimization strategy to circumvent computational burdens while identifying disease-associated risk factors within genetic variations across the entire chromosome. We assessed S-SCCA against conventional SCCA using a neuroimaging genetic dataset from the Alzheimer’s Disease Neuroimaging Initiative. Results showed that S-SCCA demonstrated comparable or superior modeling performance and feature selection capabilities. Furthermore, we applied S-SCCA to two summary statistics datasets from two large GWAS, where original imaging and genetic data were inaccessible. S-SCCA replicated the genetic loci identified by GWAS and additional meaningful variants. Additionally, it revealed bi-multivariate relationships between imaging QTs and SNPs, indicating its powerful modeling capability. These findings highlight the promise of S-SCCA as a practical bi-multivariate learning technique in brain imaging genetics, circumventing the need for sensitive individual-level imaging and genetic data, thereby enhancing its potential for broader applicability and accessibility in biomedical studies.

Applying Neural Radiance Fields (NeRF) to downstream perception tasks for scene understanding and representation is becoming increasingly popular. Most existing methods treat semantic prediction as an additional rendering task, i.e., the “label rendering” task, to build semantic NeRFs. However, by rendering semantic/instance labels per pixel without considering the contextual information of the rendered image, these methods usually suffer from unclear boundary segmentation and abnormal segmentation of pixels within an object. To solve this problem, we propose Generalized Perception NeRF (GP-NeRF), a novel pipeline that makes the widely used segmentation model and NeRF work compatibly under a unified framework, for facilitating context-aware 3D scene perception. To accomplish this goal, we introduce transformers to aggregate radiance as well as semantic embedding fields jointly for novel views and facilitate the joint volumetric rendering of both fields. In addition, we propose two self-distillation mechanisms, i.e., the Semantic Distill Loss and the Depth-Guided Semantic Distill Loss, to enhance the discrimination and quality of the semantic field and the maintenance of geometric consistency. In evaluation, as shown in Fig. 1 we conduct experimental comparisons under two perception tasks (i.e. semantic and instance segmentation) using both synthetic and real-world datasets. Notably, our method outperforms SOTA approaches by 6.94%,11.76%, and 8.47% on generalized semantic segmentation, finetuning semantic segmentation, and instance segmentation, respectively. Project.

Salient object detection (SOD) and camouflaged object detection (COD) are related yet distinct binary mapping tasks. These tasks involve multiple modalities, sharing commonalities and unique cues. Existing research often employs intricate task-specific specialist models, potentially leading to redundancy and suboptimal results. We introduce VS-Code, a generalist model with novel 2D prompt learning, to jointly address four SOD tasks and three COD tasks. We utilize VST as the foundation model and introduce 2D prompts within the encoder-decoder architecture to learn domain and task-specific knowledge on two separate dimensions. A prompt discrimination loss helps disentangle peculiarities to benefit model optimization. VSCode outperforms state-of-the-art methods across six tasks on 26 datasets and exhibits zero-shot generalization to unseen tasks by combining 2D prompts, such as RGB-D COD. Source code has been available at https://github.com/Sssssuperior/VSCode.

Existing few-shot semantic segmentation methods typically rely on a one-way flow of category information from support to query, ignoring the impact of intra-class diversity. To address this, drawing inspiration from cybernetics, we introduce a Query Feedback Branch (QFB) to propagate query information back to support, generating a query-related support prototype that is more aligned with the query. Subsequently, a Query Amplifier Branch (QAB) is employed to amplify target objects in the query using the acquired support prototype. To further improve the model, we propose a Query Rectification Module (QRM), which utilizes the prediction disparity in the query before and after support activation to identify challenging positive and negative samples from ambiguous regions for query self-rectification. Furthermore, we integrate the QFB, QAB, and QRM into a feedback and rectification layer and incorporate it into an iterative pipeline. This configuration enables the progressive enhancement of bidirectional reciprocative flow of category information between query and support, effectively providing query-adaptive support information and addressing the intra-class diversity problem. Extensive experiments conducted on both PASCAL-5i and COCO-20i datasets validate the effectiveness of our approach. The code is available at https://github.com/LIUYUANWEI98/IFRNet .

Previous knowledge distillation (KD) methods mostly focus on compressing network architectures, which is not thorough enough in deployment as some costs like transmission bandwidth and imaging equipment are related to the image size. Therefore, we propose Pixel Distillation that extends knowledge distillation into the input level while simultaneously breaking architecture constraints. Such a scheme can achieve flexible cost control for deployment, as it allows the system to adjust both network architecture and image quality according to the overall requirement of resources. Specifically, we first propose an input spatial representation distillation (ISRD) mechanism to transfer spatial knowledge from large images to student's input module, which can facilitate stable knowledge transfer between CNN and ViT. Then, a Teacher-Assistant-Student (TAS) framework is further established to disentangle pixel distillation into the model compression stage and input compression stage, which significantly reduces the overall complexity of pixel distillation and the difficulty of distilling intermediate knowledge. Finally, we adapt pixel distillation to object detection via an aligned feature for preservation (AFP) strategy for TAS, which aligns output dimensions of detectors at each stage by manipulating features and anchors of the assistant. Comprehensive experiments on image classification and object detection demonstrate the effectiveness of our method.

This paper proposes a scribble-based weakly supervised RGB-D salient object detection (SOD) method to relieve the annotation burden from pixel-wise annotations. In view of the ensuing performance drop, we summarize two natural deficiencies of the scribbles and try to alleviate them, which are the weak richness of the pixel training samples (WRPS) and the poor structural integrity of the salient objects (PSIO). WRPS hinders robust saliency perception learning, which can be alleviated via model design for robust feature learning and pseudo labels generation for training sample enrichment. Specifically, we first design a dynamic searching process module as a meta operation to conduct multi-scale and multi-modal feature fusion for the robust RGB-D SOD model construction. Then, a dual-branch consistency learning mechanism is proposed to generate enough pixel training samples for robust saliency perception learning. PSIO makes direct structural learning infeasible since scribbles can not provide integral structural supervision. Thus, we propose an edge-region structure-refinement loss to recover the structural information and make precise segmentation. We deploy all components and conduct ablation studies on two baselines to validate their effectiveness and generalizability. Experimental results on eight datasets show that our method outperforms other scribble-based SOD models and achieves comparable performance with fully supervised state-of-the-art methods.

While previous CNN-based models have exhibited promising results for salient object detection (SOD), their ability to explore global long-range dependencies is restricted. Our previous work, the Visual Saliency Transformer (VST), addressed this constraint from a transformer-based sequence-to-sequence perspective, to unify RGB and RGB-D SOD. In VST, we developed a multi-task transformer decoder that concurrently predicts saliency and boundary outcomes in a pure transformer architecture. Moreover, we introduced a novel token upsampling method called reverse T2T for predicting a high-resolution saliency map effortlessly within transformer-based structures. Building upon the VST model, we further propose an efficient and stronger VST version in this work,i.e.VST++. To mitigate the computational costs of the VST model, we propose a Select-Integrate Attention (SIA) module, partitioning foreground into fine-grained segments and aggregating background information into a single coarse-grained token. To incorporate 3D depth information with low cost, we design a novel depth position encoding method tailored for depth maps. Furthermore, we introduce a token-supervised prediction loss to provide straightforward guidance for the task-related tokens. We evaluate our VST++model across various transformer-based backbones on RGB, RGB-D, and RGB-T SOD benchmark datasets. Experimental results show that our model outperforms existing methods while achieving a 25% reduction in computational costs without significant performance compromise. The demonstrated strong ability for generalization, enhanced performance, and heightened efficiency of our VST++model highlight its potential.

When given a group of relevant images for co-salient object detection (Co-SOD), humans first summarize consensus cues from the whole group and then search for co-salient objects in each image. Most previous methods do not consider robustness, scalability, or stability in the summarization stage and adopt a simple fusion strategy to fuse consensus and image features in the searching stage. Our work presents a novel consensus-aware dynamic convolution (CADC) model directly from the “summarize and search” perspective to explicitly and effectively perform Co-SOD. For the summarization stage, we extract robust individual image features by a pooling method and integrate them to generate consensus features via self-attention, thus modeling the scalability and stability. Then, we simultaneously learn two types of consensus-aware dynamic kernels, i.e., a common kernel to capture group-wise common knowledge and adaptive kernels to mine image-specific consensus cues. For the second stage, we adopt dynamic convolution to perform object searching. A novel data synthesis strategy is also developed for model training. Although CADC has obtained competitive performance, we argue that incrementally learning dynamic kernels and representations is more intuitive and natural instead of using a simultaneous scheme, thus presenting our CADC++, an extension of CADC. Concretely, we first adopt the common kernel based dynamic convolution to capture coarse common cues as priors and then use the adaptive kernel based dynamic convolution for mining image-specific details. We also propose a recursive guidance strategy to further explore deep interactions among the two kinds of kernels and image features. Besides, we annotate several challenging attributes for Co-SOD datasets and perform attribute-based evaluation and robustness analysis to promote thorough model evaluation for the Co-SOD field. Extensive experimental results on four benchmark datasets verify both the effectiveness and robustness of our proposed method.

Using brain imaging quantitative traits (QTs) for identifying genetic risk factors is an important research topic in brain imaging genetics. Many efforts have been made for this task via building linear models between imaging QTs and genetic factors such as single nucleotide polymorphisms (SNPs). To the best of our knowledge, linear models could not fully uncover the complicated relationship due to the loci's elusive and diverse influences on imaging QTs. In this paper, we propose a novel multi-task deep feature selection (MTDFS) method for brain imaging genetics. MTDFS first builds a multi-task deep neural network to model the complicated associations between imaging QTs and SNPs. And then designs a multi-task one-to-one layer and imposes a combined penalty to identify SNPs that make significant contributions. MTDFS can not only extract the nonlinear relationship but also arms the deep neural network with feature selection. We compared MTDFS to multi-task linear regression (MTLR) and single-task DFS (DFS) methods on the real neuroimaging genetic data. The experimental results showed that MTDFS performed better than MTLR and DFS on the QT-SNP relationship identification and feature selection. Thus, MTDFS is powerful for identifying risk loci and could be a great supplement to brain imaging genetics.

Previous arts built an intimate link between knowledge distillation (KD) and label smoothing (LS) that they both impose regularization on the model training. In this paper, we delve deeper into investigating the hidden reason rendering KD and LS to exert distinct effects on a model’s potential ability in sequential knowledge transferring. Specifically, we observe that the distilled model typically exhibits much higher intra-class variance than the regularized one, consequentially acting as the better teacher. Then we devise two exploratory experiments and identify that sufficient intra-class variance retained by a teacher model is an implicit distillation recipe for achieving competitive student performance. The observed properties allow us to further put forth a simple yet beneficial approach that promotes intra-class diversity at the optimizing process of the teacher models to accomplish the most promising performance of KD. Extensive experiments are conducted on various image classification tasks across three distillation paradigms, demonstrating our proposed method’s effectiveness and generalization. Additionally, we offer new interpretations to receive a more in-depth cognition of the gap issues,i,e., better teacher, worse student, and the success of multi-generation self-distillation, respectively. Code will be made available at https://github.com/swift1988.

Targeted Attacks on Object Detection (TAOD) aim to deceive the victim detector into recognizing a specific instance as the predefined target category while minimizing the changes to the predicted bounding box of that instance. Yet, this kind of flexible attack paradigm, which is capable of manipulating the decision outcome of the victim detector, received limited attention, especially in the context of attacking object detection in optical remote sensing images, where relevant research remains a blank. To fill this gap, this paper concentrates on TAOD in optical remote sensing images, and pays attention to a fundamental question, how to deploy TAOD via the raw predictions (the predictions before non-maximum suppression) of a victim detector. In this regard, we depart from widely adopted task-independent importance measurements and hard-weighted ensemble optimization schemes present in existing methods. Instead, we first define the task-specific importance score, which considers both the qualities and the attack costs of predictions. Further, we propose the Task-Specific Importance-Aware Candidate Predictions Selection Scheme (TSIA-CPSS) alongside the Soft-Weighted Ensemble Optimization Scheme (SW-EOS). A total of eleven detectors on DIOR and DOTA, two commonly employed benchmarks, are included to comprehensively evaluate our approach. Furthermore, we indicate that the effectiveness of our approach is not only substantial for vanilla TAOD, but also can be better generalized to extended scenarios, which encompasses random TAOD, TAOD on oriented object detection, and targeted patch attacks, highlighting the noteworthy potential of our approach. Our codes will be released on Github.

This article studies change detection within pairs of optical images remotely sensed from overhead views. We consider that a high-performance solution to this task entails highly effective multi-level feature interaction. With that in mind, we propose a novel approach characterized by two attentive feature aggregation schemes that handle cross-level features in different processes. For the Siamese-based feature extraction of the bi-temporal image pair, we attach emphasis on constructing semantically strong and contextually rich pyramidal feature representations to enable comprehensive matching and differencing. To this end, we leverage a feature pyramid network and re-formulate its cross-level feature merging procedure as top-down modulation with multiplicative channel attention and additive gated attention. For the multi-level difference feature fusion, we progressively fuse the derived difference feature pyramid in an attend-then-filter manner. This makes the high-level fused features and the adjacent lower-level difference features constrain each other, and thus allows steady feature fusion for specifying change regions. In addition, we build an upsampling head as a replacement for the normal heads followed by static upsampling. Our implementation contains a stack of upsampling modules that allocate features for each pixel. Each has a learnable branch that produces attentive residuals for refining the statically upsampled results. We conduct extensive experiments on four public datasets and results show that our approach achieves state-of-the-art performance. Code is available at https://github.com/xingronaldo/CLAFA.

In recent years, few-shot object detection (FSOD) in remote sensing images has attracted increasing attention. Numerous studies address the challenges posed by both intra-class and inter-class variance through strategies such as augmenting sample diversity and incorporating multi-scale features. However, these features still encompass a considerable amount of noise attributes due to the complex characteristic of satellite images, persistently and adversely affecting classification. In contrast, we advocate for the belief that a limited yet refined set of features surpasses a multitude of coarse features. Accordingly, we tackle above issues through the meticulous refinement of representative category features, enhancing performance by eliminating irrelevant attributes that interfere with classification. Specifically, two pivotal modules: retentive compensation module (RCM) and personality filtering module (PFM), are introduced. The former module RCM systematically scrutinizes features proximate to the category center, yielding prototypes that exhibit both intra-class compactness and inter-class distinctiveness. Furthermore, the latter module PFM utilizes previous obtained prototypes to supervise the filtering process, diminishing the intra-class variance by excluding personality features which could impede the classification task. The integration of the above two modules enables a holistic feature representation, capturing inherent similarities within individual classes while accentuating distinctions between classes. Experiments have been conducted on the DIOR and NWPU VHR-10.v2 datasets, and the results demonstrate that our proposed approach exceeds several state-of-the-art methods. Code is available at https://github.com/yomik-js/RP-FSOD.

Recently, Few-Shot Object Detection (FSOD) has received considerable research attention as a strategy for reducing reliance on extensively labeled bounding boxes. However, current approaches encounter significant challenges due to the intrinsic issue of incomplete annotation while building the instance-level training benchmark. In such cases, the instances with missing annotations are regarded as background, resulting in erroneous training gradients back-propagated through the detector, thereby compromising the detection performance. To mitigate this challenge, we introduce a simple and highly efficient method that can be plugged into both meta-learning-based and transfer-learning-based methods. Our method incorporates two innovative components: Confusing Proposals Separation (CPS) and Affinity-Driven Gradient Relaxation (ADGR). Specifically, CPS effectively isolates confusing negatives while ensuring the contribution of hard negatives during model fine-tuning; ADGR then adjusts their gradients based on the affinity to different category prototypes. As a result, false-negative samples are assigned lower weights than other negatives, alleviating their harmful impacts on the few-shot detector without the requirement of additional learnable parameters. Extensive experiments conducted on the PASCAL VOC and MS-COCO datasets consistently demonstrate that our method significantly outperforms both the baseline and recent FSOD methods. Furthermore, its versatility and efficiency suggest the potential to become a stronger new baseline in the field of FSOD. Code is available at https://github.com/Ybowei/UNP.

Object detection in remote sensing images has garnered significant attention due to its wide applications in real-world scenarios. However, most existing oriented object detectors still suffer from complex backgrounds and varying angles, limiting their performance to further improvement. In this paper, we propose a novel oriented detector withHierarchical mask prompting andRobust integrated regression, termed HRDet. Specifically, to cope with the first issue, we construct a hierarchical mask prompting module consisting of a semantic mask prediction branch and hierarchical Softmax technique. The former aims to isolate object instances from cluttered interferences guided by coarse box-wise masks, while the latter propagates differentiated features for adjacent layers using hierarchical attentive weights. To deal with the second issue, we strive for robust integrated regression and formulate an efficient oriented IoU loss, explicitly measuring the discrepancies of three geometric factors in oriented regression, i.e., the central point distance, side length, and angle. This innovative loss intends to overcome the problem that existing IoU-based losses are invariant during the regression of varying angles. We applied these two strategies to a simple one-stage detection pipeline, achieving a new level of trade-off between speed and accuracy. Extensive experiments on four large aerial imagery datasets, DOTA-v1.0, DOTA-v2.0, DIOR-R, and HRSC2016, demonstrate that our HRDet significantly improves the accuracy of the one-stage detector over refine-stage counterparts while maintaining the efficiency advantage. The source code will be available athttps://github.com/yanqingyao1994/HRDet.

Fine-grained remote sensing object detection aims at precisely locating objects and determining the fine-level categories. This task is exceptionally challenging due to the substantial interclass similarity, presenting difficulties in capturing discriminative features. We attribute this to the absence of essential information that can serve as supervision for the learning. This involves comprehensive visual patterns of objects and intrinsic relationships of multigranular features. In this article, we propose a novel scheme dubbed as digging into the multigranular archetype (DIMA) for fine-grained remote sensing object detection. In detail, we first design a simple yet effective frequency-aware representation supplement (FARS) mechanism learning from original images and their auxiliary frequency counterparts simultaneously. The FARS introduces high- and low-frequency representations to reinforce a range of visual cues, such as particular regions associated with the former and contours of objects related to the latter. Then, we further devise a module named hierarchical classification paradigm (HCP), which constructs the interhierarchy relationships between coarse and fine-level representations and then exploits them to guide fine-grained feature enhancement. HCP eventually selects and boosts samples that are hard to discriminate by keeping consistency in multilevels. Our method can be easily integrated into prevailing oriented object detectors and brings consistent performance improvements across these detectors. Notably, our method combined with oriented RCNN (ORCNN) achieves 44.44% (+3.62%) on the FAIR1M and 91.0% (+6.9%) on the MAR20. Moreover, thoughtful discussions about qualitative results and rich visualizations are provided to intuitively underscore the superiority of our approach. The source code is available athttps://github.com/chengjc2019/DIMA.

Recent years have witnessed the astonishing development of transformer-based paradigm in single object tracking (SOT) in generic videos. However, due to the fact that the targets of interest in satellite videos are small in size and weak in visual appearance, the advancements of transformer-based paradigm in satellite video object tracking are impeded. To alleviate this issue, a novel transformer-based recipe is proposed, which consists of a bi-direction propagation and fusion (Bi-PF) strategy and a target-aware enhancement (TAE) module. Concretely, we first adopt the Bi-PF strategy to make full use of multiscale information to generate discriminative representations of tracking targets. Then, the TAE module is employed to decouple an object query into content-aware embedding and spatial-aware embedding and produce a target prototype to help get high-quality content-aware embedding. It is worth mentioning that, different from the previous methods in satellite video tracking most of which evaluate their performance using only several videos, we conduct extensive experiments on the SatSOT dataset which consists of 105 videos. In particular, the proposed method achieves the success score of 45.6% and the precision score of 57.6%, surpassing the baseline method by 5.0% and 9.5%, respectively. The code will be released athttps://github.com/laybebe/TATrans_SVOT.

Weakly supervised object detection (WSOD) trains detectors by only weak labels, aiming to save the burden of expensive bounding box-level annotations. Most previous efforts formulate WSOD as a multiple instance learning (MIL) problem, which is prone to detect discriminative object parts and miss object instances. This article proposes an attention erasing and instance sampling (AE-IS) approach to alleviate the above problems. Concretely, we first apply an attention erasing (AE) scheme to the WSOD model to hide the most discriminative region for capturing the integral extent of the object. Then, we employ an intersection-over-union (IoU)-balanced sampling component toward mining more object instances. Moreover, an instance reweighted loss (IRL) is designed to learn a larger portion of object instances, thereby further enhancing the performance of the object detector. Experimental results demonstrate that our method significantly improves the baseline approach by great margins and achieves competitive performance with the state-of-the-art algorithms on the NWPU VHR-10.v2 (72.0% mAP, 76.1% CorLoc) and DIOR (29.1% mAP, 55.9% CorLoc) datasets. The source code will be available athttps://github.com/XuanX/AE-IS.

Oriented object detection in aerial images has made significant advancements propelled by well-developed detection frameworks and diverse representation approaches to oriented bounding boxes. However, within modern oriented object detectors, the insufficient consideration given to certain factors, like contextual priors in aerial images and the sensitivity of the angle regression, hinder further improvement of detection performance. In this paper, we propose a dual-focused detector (DFDet), which simultaneously focuses on the exploration of contextual knowledge and the mitigation of angle sensitivity. Specifically, DFDet contains two novel designs: a contextual dependence mining network (CDMN) and a penalty-incentive allocation strategy (PIAS). CDMN constructs multiple features containing contexts across various ranges with low computational burden, and aggregates them into a compact yet informative representation that empowers the model for robust inference. PIAS dynamically calibrates the angle regression loss with a scalable penalty term determined by the angle regression sensitivity, incentivizing model to boost regression capacity for large aspect ratio objects challenging to be localized accurately. Extensive experiments on four widely-used benchmarks demonstrate the effectiveness of our approach, and new state-of-the-arts for one-stage object detection in aerial images are established. Without bells and whistles, DFDet with ResNet50 achieves 74.71% mAP running at 23.4 FPS on the most widely-used DOTA-v1.0 dataset. The source code is available at https://github.com/DDGRCF/DFDet.

Gaze estimation is an important fundamental task in computer vision and medical research. Existing works have explored various effective paradigms and modules for precisely predicting eye gazes. However, the uncertainty for gaze estimation, e.g., input uncertainty and annotation uncertainty, have been neglected in previous research. Existing models use a deterministic function to estimate the gaze, which cannot reflect the actual situation in gaze estimation. To address this issue, we propose a probabilistic framework for gaze estimation by modeling the input uncertainty and annotation uncertainty. We first utilize probabilistic embeddings to model the input uncertainty, representing the input image as a Gaussian distribution in the embedding space. Based on the input uncertainty modeling, we give an instance-wise uncertainty estimation to measure the confidence of prediction results, which is critical in practical applications. Then, we propose a new label distribution learning method, probabilistic annotations, to model the annotation uncertainty, representing the raw hard labels as Gaussian distributions. In addition, we develop an Embedding Distribution Smoothing (EDS) module and a hard example mining method to improve the consistency between embedding distribution and label distribution. We conduct extensive experiments, demonstrating that the proposed approach achieves significant improvements over baseline and state-of-the-art methods on two widely used benchmark datasets, GazeCapture and MPIIFaceGaze, as well as our collected dataset using mobile devices.

Accurate genotyping of the epidermal growth factor receptor (EGFR) is critical for the treatment planning of lung adenocarcinoma. Currently, clinical identification of EGFR genotyping highly relies on biopsy and sequence testing which is invasive and complicated. Recent advancements in the integration of computed tomography (CT) imagery with deep learning techniques have yielded a non-invasive and straightforward way for identifying EGFR profiles. However, there are still many limitations for further exploration: 1) most of these methods still require physicians to annotate tumor boundaries, which are time-consuming and prone to subjective errors; 2) most of the existing methods are simply borrowed from computer vision field which does not sufficiently exploit the multi-level features for final prediction. To solve these problems, we propose a Denseformer framework to identify EGFR mutation status in a real end-to-end fashion directly from 3D lung CT images. Specifically, we take the 3D whole-lung CT images as the input of the neural network model without manually labeling the lung nodules. This is inspired by the medical report that the mutational status of EGFR is associated not only with the local tumor nodules but also with the microenvironment surrounded by the whole lung. Besides, we design a novel Denseformer network to fully explore the distinctive information across the different level features. The Denseformer is a novel network architecture that combines the advantages of both convolutional neural network (CNN) and Transformer. Denseformer directly learns from the 3D whole-lung CT images, which preserves the spatial location information in the CT images. To further improve the model performance, we designed a combined Transformer module. This module employs the Transformer Encoder to globally integrate the information of different levels and layers and use them as the basis for the final prediction. The proposed model has been tested on a lung adenocarcinoma dataset collected at the Affiliated Hospital of Zunyi Medical University. Extensive experiments demonstrated the proposed method can effectively extract meaningful features from 3D CT images to make accurate predictions. Compared with other state-of-the-art methods, Denseformer achieves the best performance among current methods using deep learning to predict EGFR mutation status based on a single modality of CT images.

Neurodegenerative disorders usually happen stage-by-stage rather than overnight. Thus, cross-sectional brain imaging genetic methods could be insufficient to identify genetic risk factors. Repeatedly collecting imaging data over time appears to solve the problem. But most existing imaging genetic methods only use longitudinal imaging phenotypes straightforwardly, ignoring the disease progression trajectory which might be a more stable disease signature. In this paper, we propose a novel sparse multi-task mixed-effects longitudinal imaging genetic method (SMMLING). In our model, disease progression fitting and genetic risk factors identification are conducted jointly. Specifically, SMMLING models the disease progression using longitudinal imaging phenotypes, and then associates fitted disease progression with genetic variations. The baseline status and changing rate, i.e., the intercept and slope, of the progression trajectory thus shoulder the responsibility to discover loci of interest, which would have superior and stable performance. To facilitate the interpretation and stability, we employ$\ell _{{2},{1}}$-norm and the fused group lasso (FGL) penalty to identify loci at both the individual level and group level. SMMLING can be solved by an efficient optimization algorithm which is guaranteed to converge to the global optimum. We evaluate SMMLING on synthetic data and real longitudinal neuroimaging genetic data. Both results show that, compared to existing longitudinal methods, SMMLING can not only decrease the modeling error but also identify more accurate and relevant genetic factors. Most risk loci reported by SMMLING are missed by comparison methods, implicating its superiority in genetic risk factors identification. Consequently, SMMLING could be a promising computational method for longitudinal imaging genetics.

Medical report generation is a valuable and challenging task, which automatically generates accurate and fluent diagnostic reports for medical images, reducing workload of radiologists and improving efficiency of disease diagnosis. Fine-grained alignment of medical images and reports facilitates the exploration of close correlations between images and texts, which is crucial for cross-modal generation. However, visual and linguistic biases caused by radiologists' writing styles make cross-modal image-text alignment difficult. To alleviate visual-linguistic bias, this paper discretizes medical reports and introduces an intermediate modality, i.e. phrasebook, consisting of key noun phrases. As discretized representation of medical reports, phrasebook contains both disease-related medical terms, and synonymous phrases representing different writing styles which can identify synonymous sentences, thereby promoting fine-grained alignment between images and reports. In this paper, an augmented two-stage medical report generation model with phrasebook (PhraseAug) is developed, which combines medical images, clinical histories and writing styles to generate diagnostic reports. In the first stage, phrasebook is used to extract semantically relevant important features and predict key phrases contained in the report. In the second stage, medical reports are generated according to the predicted key phrases which contain synonymous phrases, promoting our model to adapt to different writing styles and generating diverse medical reports. Experimental results on two public datasets, IU-Xray and MIMIC-CXR, demonstrate that our proposed PhraseAug outperforms state-of-the-art baselines.

Deep reinforcement learning (DRL) has demonstrated impressive performance in medical image segmentation, particularly for low-contrast and small medical objects. However, current DRL-based segmentation methods face limitations due to the optimization of error propagation in two separate stages and the need for a significant amount of labeled data. In this paper, we propose a novel deep generative adversarial reinforcement learning (DGARL) approach that, for the first time, enables end-to-end semi-supervised medical image segmentation in the DRL domain. DGARL ingeniously establishes a pipeline that integrates DRL and generative adversarial networks (GANs) to optimize both detection and segmentation tasks holistically while mutually enhancing each other. Specifically, DGARL introduces two innovative components to facilitate this integration in semi-supervised settings. First, a task-joint GAN with two discriminators links the detection results to the GAN's segmentation performance evaluation, allowing simultaneous joint evaluation and feedback. This ensures that DRL and GAN can be directly optimized based on each other's results. Second, a bidirectional exploration DRL integrates backward exploration and forward exploration to ensure the DRL agent explores the correct direction when forward exploration is disabled due to lack of explicit rewards. This mitigates the issue of unlabeled data being unable to provide rewards and rendering DRL unexplorable. Comprehensive experiments on three generalization datasets, comprising a total of 640 patients, demonstrate that our novel DGARL achieves 85.02% Dice and improves at least 1.91% for brain tumors, achieves 73.18% Dice and improves at least 4.28% for liver tumors, and achieves 70.85% Dice and improves at least 2.73% for pancreas compared to the ten most recent advanced methods, our results attest to the superiority of DGARL. Code is available at GitHub.

In real-world applications, image degeneration caused by adverse weather is always complex and changes with different weather conditions from days and seasons. Systems in real-world environments constantly encounter adverse weather conditions that are not previously observed. Therefore, it practically requires adverse weather removal models to continually learn from incrementally collected data reflecting various degeneration types. Existing adverse weather removal approaches, for either single or multiple adverse weathers, are mainly designed for a static learning paradigm, which assumes that the data of all types of degenerations to handle can be finely collected at one time before a single-phase learning process. They thus cannot directly handle the incremental learning requirements. To address this issue, we made the earliest effort to investigate the continual all-in-one adverse weather removal task, in a setting closer to real-world applications. Specifically, we develop a novel continual learning framework with effective knowledge replay (KR) on a unified network structure. Equipped with a principal component projection and an effective knowledge distillation mechanism, the proposed KR techniques are tailored for the all-in-one weather removal task. It considers the characteristics of the image restoration task with multiple degenerations in continual learning, and the knowledge for different degenerations can be shared and accumulated in the unified network structure. Extensive experimental results demonstrate the effectiveness of the proposed method to deal with this challenging task, which performs competitively to existing dedicated or joint training image restoration methods. Our code is available athttps://github.com/xiaojihh/CL_all-in-one.

Besides combining appearance and motion information, another crucial factor for video salient object detection (VSOD) is to mine spatial-temporal (ST) knowledge, including complementary long-short temporal cues and global-local spatial context from neighboring frames. However, the existing methods only explored part of them and ignored their complementarity. In this article, we propose a novel complementary ST transformer (CoSTFormer) for VSOD, which has a short-global branch and a long-local branch to aggregate complementary ST contexts. The former integrates the global context from the neighboring two frames using dense pairwise attention, while the latter is designed to fuse long-term temporal information from more consecutive frames with local attention windows. In this way, we decompose the ST context into a short-global part and a long-local part and leverage the powerful transformer to model the context relationship and learn their complementarity. To solve the contradiction between local window attention and object motion, we propose a novel flow-guided window attention (FGWA) mechanism to align the attention windows with object and camera movements. Furthermore, we deploy CoSTFormer on fused appearance and motion features, thus enabling the effective combination of all three VSOD factors. Besides, we present a pseudo video generation method to synthesize sufficient video clips from static images for training ST saliency models. Extensive experiments have verified the effectiveness of our method and illustrated that we achieve new state-of-the-art results on several benchmark datasets.

With the goal of learning to localize specific object semantics using the low-cost image-level annotation, weakly supervised object localization (WSOL) has been receiving increasing attention in recent years. Although existing literatures have studied a number of major issues in this field, one important yet challenging scenario, where the test object semantics may appear in the training phase (seen categories) or never been observed before (unseen categories), is still beyond the exploration of the existing works. We define this scenario as the generalized WSOL (GWSOL) and make a pioneering effort to study it in this article. By leveraging attribute vectors to associate seen and unseen categories, we involve threefold modeling components, i.e., the class-sensitive modeling, semantic-agnostic modeling, and content-aware modeling, into a unified end-to-end learning framework. Such design enables our model to recognize and localize unconstrained object semantics, learn compact and discriminative features that could represent the potential unseen categories, and customize content-aware attribute weights to avoid localizing on misleading attribute elements. To advance this research direction, we contribute the bounding-box manual annotations to the widely used AwA2 dataset and benchmark the GWSOL methods. Comprehensive experiments demonstrate the effectiveness of our proposed learning framework and each of the considered modeling components.

Saliency prediction (SAP) plays a crucial role in simulating the visual perception function of human beings. In practical situations, humans can quickly grasp saliency extraction in new image domains. However, current SAP methods mainly concentrate on training models in single domains, which do not effectively handle diverse content and styles present in real-world images. As a result, it would be of great significance if SAP models could efficiently adjust to new image domains. To this end, this article aims to design SAP models that can imitate the incremental learning ability of human beings on multiple image domains and name domain-incremental saliency prediction (DISAP). To make a tradeoff between preventing the forgetting of historical domains and achieving high performance on new domains, we propose a progressively updated domain incremental encoder. This encoder consists of a domain-sharing branch and a domain-specific branch. The domain-sharing branch includes a feature selection mechanism to preserve crucial parameters after fine-tuning the model on each current domain. The remaining parameters are reserved to absorb knowledge from future domains. Furthermore, to capture the unique characteristics of each domain with relatively low computational overhead, we introduce a lightweight design to construct the domain-specific branch, enabling effective adaptation to new domains. Extensive experiments are conducted on multiple domain-incremental learning settings formed by four saliency prediction datasets, including Salicon, MIT1003, the art subset of CAT2000, and WebSal. The results demonstrate that our method outperforms existing methods significantly. The code is available at https://github.com/KaIi-github/DIL4SAP .

Visual attention prediction is linked to brain activity, cognition, and behavior. Despite the availability of brain activity features, previous studies have not fully utilized them, resulting in saliency maps predicted by models primarily based on image features that do not accurately reflect visual attention in the human brain. This inspires us to use functional Magnetic Resonance Imaging (fMRI) signals as a "brain observer" to supervise the training of developing models that integrate top-down image attention-dependent cues and supervise information from saliency maps generated from gaze movement patterns under natural stimuli. Hence, this paper presents an FMRI-Guided Time-Symmetric Joint Model to predict saliency maps from movie clips, which captures the dynamic aspects of human brain cognition and attention, enabling the combination of image features with brain features. Furthermore, we generalize the model to the MS-COCO challenge, evaluating its performance on non-movie data. Our model outperforms other brain-feature-free methods in focusing on visual attention regions of humans in both movie and non-movie datasets. Additionally, incorporating brain features improves model performance, indicating their ability to bridge the semantic gap between human cognition and visual images, allowing for more accurate capture of visual attention regions.

Over decades, neuroscience has accumulated a wealth of research results in the text modality that can be used to explore cognitive processes. Meta-analysis is a typical method that successfully establishes a link from text queries to brain activation maps using these research results, but it still relies on an ideal query environment. In practical applications, text queries used for meta-analyses may encounter issues such as semantic redundancy and ambiguity, resulting in an inaccurate mapping to brain images. On the other hand, large language models (LLMs) like ChatGPT have shown great potential in tasks such as context understanding and reasoning, displaying a high degree of consistency with human natural language. Hence, LLMs could improve the connection between text modality and neuroscience, resolving existing challenges of meta-analyses. In this study, we propose a method called Chat2Brain that combines LLMs to basic text-2-image model, known as Text2Brain, to map open-ended semantic queries to brain activation maps in data-scarce and complex query environments. By utilizing the understanding and reasoning capabilities of LLMs, the performance of the mapping model is optimized by transferring text queries to semantic queries. We demonstrate that Chat2Brain can synthesize anatomically plausible neural activation patterns for more complex tasks of text queries.

Most previous co-salient object detection works mainly focus on extracting co-salient cues via mining the consistency relations across images while ignore explicit exploration of background regions. In this paper, we propose a Discriminative co-saliency and background Mining Transformer framework (DMT) based on several economical multi-grained correlation modules to explicitly mine both co-saliency and background information and effectively model their discrimination. Specifically, we first propose a region-to-region correlation module for introducing inter-image relations to pixel-wise segmentation features while maintaining computational efficiency. Then, we use two types of pre-defined tokens to mine co-saliency and background information via our proposed contrast-induced pixel-to-token correlation and co-saliency token-to-token correlation modules. We also design a token-guided feature refinement module to enhance the discriminability of the segmentation features under the guidance of the learned tokens. We perform iterative mutual promotion for the segmentation feature extraction and token construction. Experimental results on three benchmark datasets demonstrate the effectiveness of our proposed method. The source code is available at: https://github.com/dragonlee258079/DMT.

Inspired by DETR variants, query-based end-to-end instance segmentation (QEIS) methods have recently outperformed CNN-based models on large-scale datasets. Yet they would lose efficacy when only a small amount of training data is available since it's hard for the crucial queries/kernels to learn localization and shape priors. To this end, this work offers a novel unsupervised pre-training solution for low-data regimes. Inspired by the recent success of the Prompting technique, we introduce a new pre-training method that boosts QEIS models by giving Saliency Prompt for queries/kernels. Our method contains three parts: 1) Saliency Masks Proposal is responsible for generating pseudo masks from unlabeled images based on the saliency mechanism. 2) Prompt-Kernel Matching transfers pseudo masks into prompts and injects the corresponding localization and shape priors to the best-matched kernels. 3) Kernel Supervision is applied to supply supervision at the kernel level for robust learning. From a practical perspective, our pre-training method helps QEIS models achieve a similar convergence speed and comparable performance with CNN-based models in low-data regimes. Experimental results show that our method significantly boosts several QEIS models on three datasets.11Code: https://github.com/lifuguan/saliency.prompt

Few-Shot Video Object Segmentation (FSVOS) aims to segment objects in a query video with the same category defined by a few annotated support images. However, this task was seldom explored. In this work, based on IPMT, a state-of-the-art few-shot image segmentation method that combines external support guidance information with adaptive query guidance cues, we propose to leverage multi-grained temporal guidance information for handling the temporal correlation nature of video data. We decompose the query video information into a clip prototype and a memory prototype for capturing local and long-term internal temporal guidance, respectively. Frame prototypes are further used for each frame independently to handle fine-grained adaptive guidance and enable bidirectional clip-frame prototype communication. To reduce the influence of noisy memory, we propose to leverage the structural similarity relation among different predicted regions and the support for selecting reliable memory frames. Furthermore, a new segmentation loss is also proposed to enhance the category discriminability of the learned prototypes. Experimental results demonstrate that our proposed video IPMT model significantly outperforms previous models on two benchmark datasets. Code is available at https://github.com/nankepan/VIPMT.

The past few years have witnessed the immense success of object detection, while current excellent detectors struggle on tackling size-limited instances. Concretely, the well-known challenge of low overlaps between the priors and object regions leads to a constrained sample pool for optimization, and the paucity of discriminative information further aggravates the recognition. To alleviate the aforementioned issues, we propose CFINet, a two-stage framework tailored for small object detection based on the Coarse-to-fine pipeline and Feature Imitation learning. Firstly, we introduce Coarse-to-fine RPN (CRPN) to ensure sufficient and high-quality proposals for small objects through the dynamic anchor selection strategy and cascade regression. Then, we equip the conventional detection head with a Feature Imitation (FI) branch to facilitate the region representations of size-limited instances that perplex the model in an imitation manner. Moreover, an auxiliary imitation loss following supervised contrastive learning paradigm is devised to optimize this branch. When integrated with Faster RCNN, CFINet achieves state-of-the-art performance on the large-scale small object detection benchmarks, SODA-D and SODA-A, underscoring its superiority over baseline detector and other mainstream detection approaches. Code is available at https://github.com/shaunyuan22/CFINet.

Image-text matching is a hot topic in multi-modal analysis. The existing image-text matching algorithms focus on bridging the heterogeneity gap and mapping the feature into a common space under strong alignment assumption. However, these methods have unsatisfactory performance under the weak alignment scenario, which assumes that the text contains more abstract information, and the number of entities in the text is always fewer than objects in image. This is the first time, from our knowledge, to solve the image-text matching problem from the perspective of information difference with weak alignment. In order to both narrow the cross-modal heterogeneity gap and balance the information discrepancy, we proposed an imagination network to enrich the text modality based on pre-trained framework, which is helpful for image-text matching. The imagination network utilizes reinforcement learning to enhance the semantic information for text modality, and an action refinement strategy is designed to constrain the freedom and divergence of imagination. The experiment results show the superiority and generality of the proposed framework based on two pre-trained models, CLIP and BLIP on two most frequently-used datasets MSCOCO and Flickr30K.

Conventional deep CNN-based segmentation approaches have achieved satisfactory performance in recent years, however, they are essentially big data-driven technologies and are difficult to generalize to unseen categories. Few-shot segmentation is subsequently developed to perform pertinent operations in a low-data regime. Unfortunately, due to the training paradigm and network architecture factors, existing methods are prone to overfit the targets of base categories and yield inaccurate segmentation boundaries, which impedes the research progress to some extent. In this paper, we propose a Holistic Prototype Activation (HPA) network to alleviate these problems. Its novel designs can be summarized in three aspects: 1) A training-free scheme to derive the prior representations of base categories. 2) Prototype Activation Module (PAM) that generates reliable activation maps and well-matched query features by filtering the objects of irrelevant classes with high confidence. 3) Cross-Referenced Decoder (CRD) for interacted feature reweighting and multi-level feature aggregation. Extensive experiments on standard few-shot segmentation benchmarks (PASCAL-5$^{i}$and COCO-20$^{i}$) verify the effectiveness of our method. On top of that, the superior performance on multiple extended tasks, such as weak-label segmentation, zero-shot segmentation, and video object segmentation, also illustrates its flexibility and versatility. Our code is publicly available athttps://github.com/chunbolang/HPA.

With the rise of deep convolutional neural networks, object detection has achieved prominent advances in past years. However, such prosperity could not camouflage the unsatisfactory situation of Small Object Detection (SOD), one of the notoriously challenging tasks in computer vision, owing to the poor visual appearance and noisy representation caused by the intrinsic structure of small targets. In addition, large-scale dataset for benchmarking small object detection methods remains a bottleneck. In this paper, we first conduct a thorough review of small object detection. Then, to catalyze the development of SOD, we construct two large-scale Small Object Detection dAtasets (SODA), SODA-D and SODA-A, which focus on the Driving and Aerial scenarios respectively. SODA-D includes 24828 high-quality traffic images and 278433 instances of nine categories. For SODA-A, we harvest 2513 high resolution aerial images and annotate 872069 instances over nine classes. The proposed datasets, as we know, are the first-ever attempt to large-scale benchmarks with a vast collection of exhaustively annotated instances tailored for multi-category SOD. Finally, we evaluate the performance of mainstream methods on SODA. We expect the released benchmarks could facilitate the development of SOD and spawn more breakthroughs in this field.

Weakly Supervised Object Detection (WSOD) is of increasing importance in the community of computer vision as its extensive applications and low manual cost. Most of the advanced WSOD approaches build upon an indefinite and quality-agnostic framework, leading to unstable and incomplete object detectors. This paper attributes these issues to the process of inconsistent learning for object variations and the unawareness of localization quality and constructs a novel end-to-end Invariant and Equivariant Network (IENet). It is implemented with a flexible multi-branch online refinement, to be naturally more comprehensive-perceptive against various objects. Specifically, IENet first performs label propagation from the predicted instances to their transformed ones in a progressive manner, achieving affine-invariant learning. Meanwhile, IENet also naturally utilizes rotation-equivariant learning as a pretext task and derives an instance-level rotation-equivariant branch to be aware of the localization quality. With affine-invariance learning and rotation-equivariant learning, IENet urges consistent and holistic feature learning for WSOD without additional annotations. On the challenging datasets of both natural scenes and aerial scenes, we substantially boost WSOD to new state-of-the-art performance. The codes have been released at: https://github.com/XiaoxFeng/IENet.

Empowered by large datasets, e.g., ImageNet and MS COCO, unsupervised learning on large-scale data has enabled significant advances for classification tasks. However, whether the large-scale unsupervised semantic segmentation can be achieved remains unknown. There are two major challenges: i) we need a large-scale benchmark for assessing algorithms; ii) we need to develop methods to simultaneously learn category and shape representation in an unsupervised manner. In this work, we propose a new problem of large-scale unsupervised semantic segmentation (LUSS) with a newly created benchmark dataset to help the research progress. Building on the ImageNet dataset, we propose the ImageNet-S dataset with 1.2 million training images and 50k high-quality semantic segmentation annotations for evaluation. Our benchmark has a high data diversity and a clear task objective. We also present a simple yet effective method that works surprisingly well for LUSS. In addition, we benchmark related un/weakly/fully supervised methods accordingly, identifying the challenges and possible directions of LUSS. The benchmark and source code is publicly available at https://github.com/LUSSeg.

Despite the progress made by few-shot segmentation (FSS) in low-data regimes, the generalization capability of most previous works could be fragile when countering hard query samples with seen-class objects. This paper proposes a fresh and powerful scheme to tackle such an intractable bias problem, dubbed base and meta (BAM). Concretely, we apply an auxiliary branch (base learner) to the conventional FSS framework (meta learner) to explicitly identify base-class objects, i.e., the regions that do not need to be segmented. Then, the coarse results output by these two learners in parallel are adaptively integrated to derive accurate segmentation predictions. Considering the sensitivity of meta learner, we further introduce adjustment factors to estimate the scene differences between support and query image pairs from both style and appearance perspectives, so as to facilitate the model ensemble forecasting. The remarkable performance gains on standard benchmarks (PASCAL-5$^{i}$, COCO-20$^{i}$, and FSS-1000) manifest the effectiveness, and surprisingly, our versatile scheme sets new state-of-the-arts even with two plain learners. Furthermore, in light of its unique nature, we also discuss several more practical but challenging extensions, including generalized FSS, 3D point cloud FSS, class-agnostic FSS, cross-domain FSS, weak-label FSS, and zero-shot segmentation. Our source code is available athttps://github.com/chunbolang/BAM.

Object detection is a fundamental yet challenging task in computer vision. Despite the great strides made over recent years, modern detectors may still produce unsatisfactory performance due to certain factors, such as non-universal object features and single regression manner. In this paper, we draw on the idea of mutual-assistance (MA) learning and accordingly propose a robust one-stage detector, referred as MADet, to address these weaknesses. First, the spirit of MA is manifested in the head design of the detector. Decoupled classification and regression features are reintegrated to provide shared offsets, avoiding inconsistency between feature-prediction pairs induced by zero or erroneous offsets. Second, the spirit of MA is captured in the optimization paradigm of the detector. Both anchor-based and anchor-free regression fashions are utilized jointly to boost the capability to retrieve objects with various characteristics, especially for large aspect ratios, occlusion from similar-sized objects, etc. Furthermore, we meticulously devise a quality assessment mechanism to facilitate adaptive sample selection and loss term reweighting. Extensive experiments on standard benchmarks verify the effectiveness of our approach. On MS-COCO, MADet achieves 42.5% AP with vanilla ResNet50 backbone, dramatically surpassing multiple strong baselines and setting a new state of the art.

Weakly supervised temporal action localization is a newly emerging yet widely studied topic in recent years. The existing methods can be categorized into two localization-by-classification pipelines, i.e., the pre-classification pipeline and the post-classification pipeline. The pre-classification pipeline first performs classification on each video snippet, and then, aggregates the snippet-level classification scores to obtain the video-level classification score. In contrast, the post-classification pipeline aggregates the snippet-level features first and then predicts the video-level classification score based on the aggregated feature. Although the classifiers in these two pipelines are used in different ways, the role they play is exactly the same-to classify the given features to identify the corresponding action categories. To this end, an ideal classifier can make both pipelines work. This inspires us to simultaneously learn these two pipelines in a unified framework to obtain an effective classifier. Specifically, in the proposed learning framework, we implement two parallel network streams to model the two localization-by-classification pipelines simultaneously and make the two network streams share the same classifier. This achieves the novel Equivalent Classification Mapping (ECM) mechanism. Moreover, we discover that an ideal classifier may possess two characteristics: 1) the frame-level classification scores obtained from the pre-classification stream and the feature aggregation weights in the post-classification stream should be consistent; and 2) the classification results of these two streams should be identical. Based on these two characteristics, we further introduce a weight-transition module and an equivalent training strategy into the proposed learning framework, which assists to thoroughly mine the equivalence mechanism. Comprehensive experiments are conducted on three benchmarks and ECM achieves accurate action localization results.

Deep cross-modal hashing has achieved excellent retrieval performance with the powerful representation capability of deep neural networks. Regrettably, current methods are inevitably vulnerable to adversarial attacks, especially well-designed subtle perturbations that can easily fool deep cross-modal hashing models into returning irrelevant or the attacker’s specified results. Although adversarial attacks have attracted increasing attention, there are few studies on specialized attacks against deep cross-modal hashing. To solve these issues, we propose a targeted adversarial attack method against deep cross-modal hashing retrieval in this paper. To the best of our knowledge, this is the first work in this research field. Concretely, we first build a progressive fusion module to extract fine-grained target semantics through a progressive attention mechanism. Meanwhile, we design a semantic adaptation network to generate the target prototype code and reconstruct the category label, thus realizing the semantic interaction between the target semantics and the implicit semantics of the attacked model. To bridge modality gaps and preserve local example details, a semantic translator seamlessly translates the target semantics and then embeds them into benign examples in collaboration with a U-Net framework. Moreover, we construct a discriminator for adversarial training, which enhances the visual realism and category discrimination of adversarial examples, thus improving their targeted attack performance. Extensive experiments on widely tested cross-modal retrieval datasets demonstrate the superiority of our proposed method. Also, transferable attacks show that our generated adversarial examples have well generalization capability on targeted attacks. The source codes and datasets are available athttps://github.com/tswang0116/TA-DCH.

Fine-grained oriented object recognition (FGO2R) is a practical need for intellectually interpreting remote sensing images. It aims at realizing fine-grained classification and precise localization with oriented bounding boxes, simultaneously. Our considerations for the task are general but decisive: (i) the extraction of subtle differences carries a big weight in differentiating fine-grained classes, and (ii) oriented localization prefers rotation-sensitive features. In this article, we propose a network with separate feature refinement (SFRNet), in which two transformer-based branches are designed to perform function-specific feature refinement for fine-grained classification and oriented localization, separately. To highlight the discriminative information advantageous to fine-grained classification, we propose a spatial and channel transformer (SC-Former) to capture both the long-range spatial interactions and the key correlations hidden in the feature channels. Besides, we design a Multi-RoI loss (MRL) following the protocol of deep metric learning to enhance the separability of fine-grained classes further. For oriented localization, we integrate the oriented response convolution with the transformer structure (namely, OR-Former) to assist in encoding rotation information during regression. Extensive experimental results validate the effectiveness and robustness of our SFRNet. Without bells and whistles, our SFRNet achieves state-of-the-art performance on the large-scale FAIR1M datasets (FAIR1M-1.0 and FAIR1M-2.0). Code will be available at https://github.com/Ranchosky/SFRNet.

Deep learning-based synthetic aperture radar (SAR) target detection has been developed for years, with many advanced methods proposed to achieve higher indicators of accuracy and speed. In spite of this, the current deep detectors cannot express the reliability and interpretation in trusting the predictions, which are crucial especially for those ordinary users without much expertise in understanding SAR images. To achieve explainable SAR target detection, it is necessary to answer the following questions: how much should we trust and why cannot we trust the results. With this purpose, we explore the uncertainty for SAR target detection in this article by quantifying the model uncertainty and explaining the ignorance of the detector. First, the Bayesian deep detectors (BDDs) are constructed for uncertainty quantification, answering how much to trust the classification and localization result. Second, an occlusion-based explanation method (U-RISE) for BDD is proposed to account for the SAR scattering features that cause uncertainty or promote trustworthiness. We introduce the probability-based detection quality (PDQ) and multielement decision space for evaluation besides the traditional metrics. The experimental results show that the proposed BDD outperforms the counterpart frequentist object detector, and the output probabilistic results successfully convey the model uncertainty and contribute to more comprehensive decision-making. Furthermore, the proposed U-RISE generates an attribution map with intuitive explanations to reveal the complex scattering phenomena about which BDD is uncertain. We deem our work will facilitate explainable and trustworthy modeling in the field of SAR image understanding and increase user comprehension of model decisions.

Few-shot segmentation (FSS), which aims to determine specific objects in the query image given only a handful of densely labeled samples, has received extensive academic attention in recent years. However, most existing FSS methods are designed for natural images, and few works have been done to investigate more realistic and challenging applications,e.g., remote sensing image understanding. In such a setup, the complex nature of the raw images would undoubtedly further increase the difficulty of the segmentation task. To couple with potential inference failures, we propose a novel and powerful remote sensing FSS framework with global Rectification and decoupled Registration, termed R2Net. Specifically, a series of dynamically updated global prototypes are utilized to provide auxiliary non-target segmentation cues and to prevent inaccurate prototype activation resulting from the variability between query-support image pairs. The foreground and background information flows are then decoupled for more targeted and tailored object localization, avoiding unnecessary confusion from information redundancy. Furthermore, we impose additional constraints to promote the interclass separability and intraclass compactness. Extensive experiments on the standard benchmark iSAID-5idemonstrate the superiority of the proposed R2Net over state-of-the-art FSS models. The code will be made available.

In recent years, few-shot segmentation (FSS) has received widespread attention from scholars by virtue of its superiority in low-data regimes. Most existing research focuses on natural image processing, and very few studies are dedicated to the practical but challenging topic of remote sensing image understanding. Related experimental results show that directly transferring the previously proposed framework to the current domain is prone to produce unsatisfactory results withincomplete objectsandirrelevant distractors. Such phenomena can be attributed to the lack of modules specifically designed for the complex characteristics of remote sensing images,e.g., great intra-class diversity and low target-background contrast. In this paper, we propose a conceptually simple and easy-to-implement framework to tackle the aforementioned problems. Specifically, our innovative design embodies two main aspects: i) the support mask is progressively parsed into multiple valuable sub-regions that can be further exploited to compute local descriptors with segmentation cues about intractable parts; ii) the base-class memories stored in the meta-training phase are replayed and leveraged for the distillation of novel-class prototypes, where the commonalities between classes are adequately explored, more in line with the concept oflearning to learn. These two components, i.e., the progressive parsing module and commonality distillation module, contribute to each other and together constitute the proposed PCNet. We conduct extensive experiments on the standard benchmark to evaluate segmentation performance in few-shot settings. Quantitative and qualitative results illustrate that our PCNet distinctly outperforms previous FSS approaches and sets a new state-of-the-art.

Practical applications ask for object detection models that achieve high performance at low overhead. Knowledge distillation demonstrates favorable potential in this case by transferring knowledge from a cumbersome teacher model to a lightweight student model. However, previous distillation methods are plagued with massive misleading background information in remote sensing images and ignore investigating the relationships between different instances. In this article, we propose an instance-aware distillation (InsDist for short) method to derive efficient remote sensing object detectors. Our InsDist combines feature-based and relation-based knowledge distillation to make the most of instance-related information in the knowledge transfer from the teacher to the student. On one hand, we propose a parameter-free masking module to decouple instance-related foreground from instance-irrelevant background in multiscale features. On the other hand, we construct the relationships between different instances to enhance the learning of intraclass compactness and interclass dispersion. The student comprehensively imitates both features and relationships from the teacher, yielding considerable effectiveness in dealing with complex remote sensing images. In addition, our InsDist can be easily built on mainstream object detectors with negligible extra cost. Extensive experiments on two large-scale remote sensing object detection datasets, namely DIOR and DOTA, show that our InsDist obtains noticeable gains over other distillation methods for both one-stage and two-stage, as well as both anchor-based and anchor-free detectors. The source code will be publicly available athttps://github.com/swift1988/InsDist.

Few-shot aerial image object detection aims to rapidly detect object instances of novel category in aerial images by using few labeled samples. However, due to the complex background of aerial images, few labeled samples of novel categories, and the model trained with the few-shot learning paradigm is biased towards the base categories, it greatly increases the difficulty of identifying foreground objects of novel categories. In addition to this, tiny object detection is always a hot potato in aerial image object detection, and it is even more difficult for few-shot object detection. To this end, we propose a Few-shot aerial image object detection with Confidence-Iou collaborative proposal filtration and Tiny object constraint loss (FsCIT). Specifically, we first introduce a new confidence-iou collaborative proposal filtration scheme to RPN, which combines the unbiased IoU scores between the two bounding boxes with foreground-background confidence scores to filter redundant region proposals and rescue more foreground proposals for the novel categories in RPN. Then, we design a new tiny object loss constraint term to attempt at overcoming the challenge of tiny object detection in few-shot aerial image object detection. This term considers the central point distance, the size of ground-truth bounding boxes, and the distances between the four edges of the ground-truth bounding box and the predicted bounding box. Experiments on DIOR, AI-TOD and HRRSD datasets show that FsCIT is effective and can improve the performance of few-shot aerial image object detection.

Oriented object detection (OOD) aims to precisely detect the objects with arbitrary orientation in remote sensing images. Up to now, most of bounding box regression (BBR) losses for OOD are transferred from horizontal object detection (HOD) methods, however, the transferring requires lots of professional knowledge and experiences for designers, consequently, many excellent BBR losses for HOD have not been transferred to OOD. To accelerate the research progress of BBR loss for OOD, a unified transferring strategy (UTS) is proposed to facilitate the transferring of BBR loss from HOD to OOD. The UTS proposes that the BBR of oriented bounding box (OBB) can be converted into the joint BBR of its horizontal smallest enclosing rectangle (HSER) and two offsets, so the BBR loss in HOD can be easily transferred to OOD by using HSER as a bridge. Following the UTS, a BBR loss named Rotated-IoU (RIoU) loss is designed for OOD by transferring an advanced BBR loss in HOD, which can be considered as an example to show how to transfer. On the basis of RIoU loss, a focal rotated-IoU (FRIoU) loss is proposed to assign larger weights to hard samples in the BBR. The comparisons with other BBR losses show that the RIoU and FRIoU losses can give better performance. The ablation study shows that giving more attention to hard samples in BBR is effective. The comparisons with many advanced methods demonstrate that the combinations of baseline methods and FRIoU loss achieve state-of-the-art performance on the DOTA and DIOR-R datasets.

Oriented object detectors have suffered severely from the discontinuous boundary problem for a long time. In this work, we ingeniously avoid this problem by relating regression outputs to regression target orientations. The core idea of our method is to build a contour function which imports orientations and outputs the corresponding distance predictions. Inspired by Fourier transformations, we assume this function can be represented as a linear combination of trigonometric functions and Fourier series. We replace the final 4D layer in the regression branch of fully convolutional one-stage object detector (FCOS) with a Fourier Series Transformation (FST) module and term this new network FCOSF. By this unique design, the regression outputs in FCOSF can adaptively vary according to the regression target orientations. Thus, the discontinuous boundary has no impact on our FCOSF. More importantly, FCOSF avoids building complicated oriented box representations, which usually cause extra computations and ambiguities. With only flipping augmentation and single-scale training and testing, FCOSF with ResNet-50 achieves 73.64% mAP on the DOTA-v1.0 dataset with up to 23.6 FPS speed, surpassing all one-stage oriented object detectors. On the more challenging DOTA-v2.0 dataset, FCOSF also achieves the highest results of 51.75% mAP among one-stage detectors. More experiments on DIOR-R and HRSC2016 are also conducted to verify the robustness of FCOSF. Code and models will be available at https://github.com/DDGRCF/FCOSF.

Advanced Patch Attacks (PAs) on object detection in natural images have pointed out the great safety vulnerability in methods based on deep neural networks. However, little attention has been paid to this topic in Optical Remote Sensing Images (O-RSIs). To this end, we focus on this research,i.e., PAs on object detection in O-RSIs, and propose a more Threatening PA without the scarification of the visual quality, dubbed TPA. Specifically, to address the problem of inconsistency between local and global landscapes in existing patch selection schemes, we propose leveraging the First-Order Difference (FOD) of the objective function before and after masking to select the sub-patches to be attacked. Further, considering the problem of gradient inundation when applying existing coordinate-based loss to PAs directly, we design an IoU-based objective function specific for PAs, dubbed Bounding box Drifting Loss (BDL), which pushes the detected bounding boxes far from the initial ones until there are no intersections between them. Finally, on two widely used benchmarks,i.e., DIOR and DOTA, comprehensive evaluations of our TPA with four typical detectors (Faster R-CNN, FCOS, RetinaNet, and YOLO-v4) witness its remarkable effectiveness. To the best of our knowledge, this is the first attempt to study the PAs on object detection in O-RSIs, and we hope this work can get our readers interested in studying this topic.

Detecting objects in remote sensing images (RSIs) using oriented bounding boxes (OBBs) is flourishing but challenging, wherein the design of OBB representations is the key to achieving accurate detection. In this article, we focus on two issues that hinder the performance of the two-stage oriented detectors: 1) the notorious boundary discontinuity problem, which would result in significant loss increases in boundary conditions, and 2) the inconsistency in regression schemes between the two stages. We propose a simple and effective bounding box representation by drawing inspiration from the polar coordinate system and integrate it into two detection stages to circumvent the two issues. The first stage specifically initializes four quadrant points as the starting points of the regression for producing high-quality oriented candidates without any postprocessing. In the second stage, the final localization results are refined using the proposed novel bounding box representation, which can fully release the capabilities of the oriented detectors. Such consistency brings a good trade-off between accuracy and speed. With only flipping augmentation and single-scale training and testing, our approach with ResNet-50-FPN harvests 76.25% mAP on the DOTA dataset with a speed of up to 16.5 frames/s, achieving the best accuracy and the fastest speed among the mainstream two-stage oriented detectors. Additional results on the DIOR-R and HRSC2016 datasets also demonstrate the effectiveness and robustness of our method. The source code is publicly available athttps://github.com/yanqingyao1994/QPDet.

Known as a hard nut, the single-model transferable targeted attacks via decision-level optimization objectives have attracted much attention among scholars for a long time. On this topic, recent works devoted themselves to designing new optimization objectives. In contrast, we take a closer look at the intrinsic problems in three commonly adopted optimization objectives, and propose two simple yet effective methods in this paper to mitigate these intrinsic problems. Specifically, inspired by the basic idea of adversarial learning, we, for the first time, propose a unified Adversarial Optimization Scheme (AOS) to release both the problems of gradient vanishing in cross-entropy loss and gradient amplification in Po+Trip loss, and indicate that our AOS, a simple transformation on the output logits before passing them to the objective functions, can yield considerable improvements on the targeted transferability. Besides, we make a further clarification on the preliminary conjecture in Vanilla Logit Loss (VLL) and point out the problem of unbalanced optimization in VLL, in which the source logit may risk getting increased without the explicit suppression on it, leading to the low transferability. Then, the Balanced Logit Loss (BLL) is further proposed, where we take both the source logit and the target logit into account. Comprehensive validations witness the compatibility and the effectiveness of the proposed methods across most attack frameworks, and their effectiveness can also span two tough cases (i.e., the low-ranked transfer scenario and the transfer to defense methods) and three datasets (i.e., the ImageNet, CIFAR-10, and CIFAR-100). Our source code is available at https://github.com/xuxiangsun/DLLTTAA.

An essential need for accurate visual object tracking is to capture better correlations between the tracking target and the search region. However, the dominant Siamese-based trackers are limited to producing dense similarity maps at once via a cross-correlations operation, ignoring to remedy the contamination caused by erroneous or ambiguous matches. In this paper, we propose a novel tracker, termed neighborhood consensus constraint-based siamese tracker (NCSiam), which takes the idea of neighborhood consensus constraint to refine the produced correlation maps. The intuition behind our approach is that we can support the nearby erroneous or ambiguous matches by analyzing a larger context of the scene that contains a unique match. Specifically, we devise a 4D convolution-based multi-level similarity refinement (MLSR) strategy. Taking the primary similarity maps obtained from a cross-correlation as input, MLSR acquires reliable matches by analyzing neighborhood consensus patterns in 4D space, thus enhancing the discriminability between the tracking target and the distractors. Besides, traditional Siamese-based trackers directly perform classification and regression on similarity response maps which discard appearance or semantic information. Therefore, an appearance affinity decoder (AAD) is developed to take full advantage of the semantic information of the search region. To further improve performance, we design a task-specific disentanglement (TSD) module to decouple the learned representations into classification-specific and regression-specific embeddings. Extensive experiments are conducted on six challenging benchmarks, including GOT-10k, TrackingNet, LaSOT, UAV123, OTB2015, and VOT2020. The results demonstrate the effectiveness of our method. The code will be available at https://github.com/laybebe/NCSiam.

Benefiting from advances in few-shot learning techniques, their application to dense prediction tasks (e.g., segmentation) has also made great strides in the past few years. However, most existing few-shot segmentation (FSS) approaches follow a similar pipeline to that of few-shot classification, where some core components are directly exploited regardless of various properties between tasks. We note that such an ill-conceived framework introduces unnecessary information loss, which is clearly unacceptable given the already very limited training sample. To this end, we delve into the typical types of information loss and provide a reasonably effective way, namely Retain And REcover (RARE). The main focus of this paper can be summarized as follows: (i) the loss of spatial information due to global pooling; (ii) the loss of boundary information due to mask interpolation; (iii) the degradation of representational power due to sample averaging. Accordingly, we propose a series of strategies to retain/recover the avoidable/unavoidable information, such as unidirectional pooling, error-prone region focusing, and adaptive integration. Extensive experiments on two popular benchmarks (i.e., PASCAL-5iand COCO-20i) demonstrate the effectiveness of our scheme, which is not restricted to a particular baseline approach. The ultimate goal of our work is to address different information loss problems within a unified framework, and it also exhibits superior performance compared to other methods with similar motivations. The source code will be made available at https://github.com/chunbolang/RARE.

Automatically generating the “impression” section of a radiology report given the “findings” section can summarize as much salient information of the “findings” section as possible, thus promoting more effective communication between radiologists and referring physicians. To significantly reduce the workload of radiologists, we develop and evaluate a novel framework of abstractive summarization methods to automatically generate the “impression” section of chest radiology reports. Despite recent advancements in natural language process (NLP) field such as BERT and its variants, existing abstractive summarization models and methods could not be directly applied to radiology reports, partly due to domain-specific radiology terminology. In response, we develop a pre-trained language model in the chest radiology domain, named ChestXRayBERT, to solve the problem of automatically summarizing chest radiology reports. Specifically, we first collect radiology-related scientific papers as pre-training corpus and pre-train a ChestXRayBERT on it. Then, an abstractive summarization model is proposed, which consists of the pre-trained ChestXRayBERT and a Transformer decoder. Finally, the model is fine-tuned on chest X-ray reports for the abstractive summarization task. When evaluated on the publicly available OPEN-I and MIMIC-CXR datasets, the performance of our proposed model achieves significant improvement compared with other neural networks-based abstractive summarization models. In general, the proposed ChestXRayBERT demonstrates the feasibility and promise of tailoring and extending advanced NLP techniques to the domain of medical imaging and radiology, as well as in the broader biomedicine and healthcare fields in the future.

Inspired by the powerful representation capability of deep neural networks, deep cross-modal hashing methods have recently drawn much attention and various deep cross-modal hashing methods have been developed. However, two key problems have not been solved well yet: 1) With advanced neural network models, how to seek the multi-modal alignment space which can effectively model the intrinsic multi-modal correlations and reduce the heterogeneous modality gaps. 2) How to effectively and efficiently preserve the modelled multi-modal semantic correlations into the binary hash codes under the deep learning paradigm. In this paper, we propose aHierarchical Message Aggregation Hashing(HMAH) method within an efficient teacher-student learning framework. Specifically, on the teacher end, we develop hierarchical message aggregation networks to construct a multi-modal complementary space by aggregating the semantic messages hierarchically across different modalities, which can better align the heterogeneous modalities and model the fine-grained multi-modal correlations. On the student end, we train a couple of student modules that learn hash functions to support cross-modal retrieval. We design a cross-modal correlation knowledge distillation strategy which seamlessly transfers the modelled fine-grained multi-modal semantic correlations from the teacher to the lightweight student modules. With the fine-grained knowledge supervision from teacher module, the semantic representation capability of hash functions can be enhanced. In addition, the whole learning framework avoids the time-consuming finetuning on the pre-trained deep models as existing methods and it is computationally efficient. Experimental results demonstrate the significant performance improvement of the proposed method on both retrieval accuracy and efficiency, compared with the state-of-the-art deep cross-modal hashing methods.

Occlusion is a challenging yet commonly seen problem for facial perception. Existing works resort to deep learning models and perform model training on synthesized data due to the lack of paired real-world data. As a result,they usually perform unsatisfactorily on real-world occluded faces because of domain gaps. In this paper, we decompose the face de-occlusion task into three stages, i.e., occlusion detection, face parsing, and face reconstruction, to alleviate this issue. We first perform occlusion detection and use its results as guidance for the second stage to conduct occlusion-free face parsing. As such, face de-occlusion is first performed on the face paring space with less difficulty. We can train these two stages on both synthesized and real-world images, hence can obtain accurate results for the latter. In the last stage, we use the domain-agnostic occlusion detection map and the face parsing map as the guidance to conduct face reconstruction, thus can reduce the impact of appearance information and improve the model performance on real-world data. Aiming at improving the model capacity of inferring occluded facial appearance, we also propose two types of reference modules to use relevant facial parts to enhance the reconstruction of occluded regions. Consequently, our proposed model achieves promising face de-occlusion results on real-world images.

Face parsing aims to assign pixel-wise semantic labels to different facial components (e.g., hair, brows, and lips) in given face images. However, directly predicting pixel-level labels for each facial component over the whole face image would obtain limited accuracy, especially for tiny facial components. To address this problem, some recent works propose to first crop tiny patches from the whole face image and then predict masks for each facial component. However, such cropping-and-segmenting strategy consists of two independent stages, which cannot be jointly optimized. Besides, as one valuable piece of information for parsing the highly structured facial components, context cues are not elaborately explored by the existing works. To address these issues, we propose a component-level refinement network (CLRNet) for precisely segmenting out each facial component. Specifically, we introduce an attention mechanism to bridge the two independent stages together and form an end-to-end trainable pipeline for face parsing. Furthermore, we incorporate the global context information into the refining process for each cropped facial component patch, providing informative cues for accurate parsing. Extensive experiments are carried out on two benchmark datasets, LFW-PL and HELEN. The results demonstrate the superiority of the proposed CLRNet over other state-of-the-art methods, especially for tiny facial components.

Human observers are the ultimate receivers and evaluators of the image visual information and have powerful perception ability of visual quality with short-term global perception and long-term regional observation. Thus, it is natural to design an image quality assessment (IQA) computational model to act like an observer for accurately predicting the human perception of image quality. Inspired by this, here, we propose a novel observer-like network (OLN) to perform IQA by jointly considering the global glimpsing information and local scanning information. Specifically, the OLN consists of a global distortion perception (GDP) module and a local distortion observation (LDO) module. The GDP module is designed to mimic the observer's global perception of image quality through performing classification of images' distortion categories and levels. Simultaneously, to simulate the human local observation behavior, the LDO module attempts to gather the long-term regional observation information of the distorted images by continuously tracing the human scanpath in the observer-like scanning manner. By leveraging the bilinear pooling layer to collaborate the short-term global perception with the long-term regional observation, our network precisely predicts the quality scores of distorted images, such as human observers. Comprehensive experiments on the public datasets powerfully demonstrate that the proposed OLN achieves state-of-the-art performance.

Alzheimer’s disease (AD) is a very complex neurodegenerative disease. Generally, different diagnostic groups could exhibit discriminative and specific patterns, including the single nucleotide polymorphisms (SNPs), brain imaging quantitative traits (QTs), as well as their associations, which may facilitate the comprehensive understanding of AD. However, most existing methods cannot guarantee to identify discriminative or class-specific biomarkers or both of them. To overcome this shortcoming, we propose a sparse multi-task contrastive and discriminative learning approach (MTCDA) to jointly learn the discriminative and specific patterns for multiple diagnostic groups. MTCDA can identify the class-relevant and discriminative SNP-QTs associations, and relevant SNPs, imaging QTs underpinning this relationship. We introduce an efficient algorithm to solve the proposed method which converges to a local optimum. The experimental results on Alzheimer’s Disease Neuroimaging Initiative (ADNI) show that MTCDA can obtain higher canonical correlation coefficients, classification accuracy and better feature selection results than state-of-the-art methods, which demonstrates the potential of our method for multi-class brain imaging genetics.

Due to the constraints of the imaging device and high cost in operation time, computer tomography (CT) scans are usually acquired with low within-slice resolution. Improving the inter-slice resolution is beneficial to the disease diagnosis for both human experts and computer-aided systems. To this end, this paper builds a novel medical slice synthesis to increase the inter-slice resolution. Considering that the groundtruth intermediate medical slices are always absent in clinical practice, we introduce the incremental cross-view mutual distillation strategy to accomplish this task in the self-supervised learning manner. Specifically, we model this problem from three different views: slice-wise interpolation from axial view and pixel-wise interpolation from coronal and sagittal views. Under this circumstance, the models learned from different views can distill valuable knowledge to guide the learning processes of each other. We can repeat this process to make the models synthesize intermediate slice data with increasing between-slice resolution. To demonstrate the effectiveness of the proposed approach, we conduct comprehensive experiments on a large-scale$CT$dataset. Quantitative and qualitative comparison results show that our method outperforms state-of-the-art algorithms by clear margins.

Object rotation is among longstanding, yet still unexplored, hard issues encountered in the task of weakly supervised object detection (WSOD) from aerial images. Existing predominant WSOD approaches built on regular CNNs which are not inherently designed to tackle object rotations without corresponding constraints, thereby leading to rotation-sensitive object detector. Meanwhile, current solutions have been prone to fall into the issue with unsTable detectors, as they ignore lower-scored instances and may regard them as backgrounds. To address these issues, in this paper, we construct a novel end-to-end weakly supervised Rotation-Invariant aerial object detection Network (RINet). It is implemented with a flexible multi-branch online detector refinement, to be naturally more rotation-perceptive against oriented objects. Specifically, RINet first performs label propagating from the predicted instances to their rotated ones in a progressive refinement manner. Meanwhile, we propose to couple the predicted in-stance labels among different rotation-perceptive branches for generating rotation-consistent supervision and mean-while pursuing all possible instances. With the rotation-consistent supervisions, RINet enforces and encourages consistent yet complementary feature learning for WSOD without additional annotations and hyper-parameters. On the challenging NWPU VHR-10.v2 and DIOR datasets, extensive experiments clearly demonstrate that we significantly boost existing WSOD methods to a new state-of-the-art performance. The code will be available at: https://github.com/XiaoxFeng/RINet.

Zero-shot object detection aims at incorporating class semantic vectors to realize the detection of (both seen and) unseen classes given an unconstrained test image. In this study, we reveal the core challenges in this research area: how to synthesize robust region features (for unseen objects) that are as intra-class diverse and inter-class separable as the real samples, so that strong unseen object detectors can be trained upon them. To address these challenges, we build a novel zero-shot object detection framework that contains an Intra-class Semantic Diverging component and an Inter-class Structure Preserving component. The former is used to realize the one-to-more mapping to obtain diverse visual features from each class semantic vector, preventing miss-classifying the real unseen objects as image backgrounds. While the latter is used to avoid the synthesized features too scattered to mix up the inter-class and foreground-background relationship. To demonstrate the effectiveness of the proposed approach, comprehensive experiments on PASCAL VOC, COCO, and DIOR datasets are conducted. Notably, our approach achieves the new state-of-the-art performance on PASCAL VOC and COCO and it is the first study to carry out zero-shot object detection in remote sensing imagery.

Recently few-shot segmentation (FSS) has been extensively developed. Most previous works strive to achieve generalization through the meta-learning framework derived from classification tasks; however, the trained models are biased towards the seen classes instead of being ideally class-agnostic, thus hindering the recognition of new concepts. This paper proposes a fresh and straightforward insight to alleviate the problem. Specifically, we apply an additional branch (base learner) to the conventional FSS model (meta learner) to explicitly identify the targets of base classes, i.e., the regions that do not need to be segmented. Then, the coarse results output by these two learners in parallel are adaptively integrated to yield precise segmentation prediction. Considering the sensitivity of meta learner, we further introduce an adjustment factor to estimate the scene differences between the input image pairs for facilitating the model ensemble forecasting. The substantial performance gains on PASCAL-5iand COCO-20iverify the effectiveness, and surprisingly, our versatile scheme sets a new state-of-the-art even with two plain learners. Moreover, in light of the unique nature of the proposed approach, we also extend it to a more realistic but challenging setting, i.e., generalized FSS, where the pixels of both base and novel classes are required to be determined. The source code is available at github.com/chunbolang/BAM.

Existing studies in few-shot semantic segmentation only focus on mining the target object information, however, often are hard to tell ambiguous regions, especially in non-target regions, which include background (BG) and Distracting Objects (DOs). To alleviate this problem, we propose a novel framework, namely Non-Target Region Eliminating (NTRE) network, to explicitly mine and eliminate BG and DO regions in the query. First, a BG Mining Module (BGMM) is proposed to extract the BG region via learning a general BG prototype. To this end, we design a BG loss to supervise the learning of BGMM only using the known target object segmentation ground truth. Then, a BG Eliminating Module and a DO Eliminating Module are proposed to successively filter out the BG and DO information from the query feature, based on which we can obtain a BG and DO-free target object segmentation result. Furthermore, we propose a prototypical contrastive learning algorithm to improve the model ability of distinguishing the target object from DOs. Extensive experiments on both PASCAL-5iand COCO-20idatasets show that our approach is effective despite its simplicity. Code is available at https://github.com/LIUYUANWEI98/NERTNet

Without access to the training data where a black-box victim model is deployed, training a surrogate model for black-box adversarial attack is still a struggle. In terms of data, we mainly identify three key measures for effective surrogate training in this paper. First, we show that leveraging the loss introduced in this paper to enlarge the inter-class similarity makes more sense than enlarging the inter-class diversity like existing methods. Next, unlike the approaches that expand the intra-class diversity in an implicit model-agnostic fashion, we propose a loss function specific to the surrogate model for our generator to enhance the intra-class diversity. Finally, in accordance with the in-depth observations for the methods based on proxy data, we argue that leveraging the proxy data is still an effective way for surrogate training. To this end, we propose a triple-player framework by introducing a discriminator into the traditional data-free framework. In this way, our method can be competitive when there are few semantic overlaps between the scarce proxy data (with the size between 1 k and 5k) and the training data. We evaluate our method on a range of victim models and datasets. The extensive results witness the effectiveness of our method. Our source code is available at https://github.com/xuxiangsun/ST-Data.

Online action detection has attracted increasing research interests in recent years. Current works model historical dependencies and anticipate the future to perceive the action evolution within a video segment and improve the detection accuracy. However, the existing paradigm ignores category-level modeling and does not pay sufficient attention to efficiency. Considering a category, its representative frames exhibit various characteristics. Thus, the category-level modeling can provide complimentary guidance to the temporal dependencies modeling. This paper develops an effective exemplar-consultation mechanism that first measures the similarity between a frame and exemplary frames, and then aggregates exemplary features based on the similarity weights. This is also an efficient mechanism, as both similarity measurement and feature aggregation require limited computations. Based on the exemplar-consultation mechanism, the long-term dependencies can be captured by regarding historical frames as exemplars, while the category-level modeling can be achieved by regarding representative frames from a category as exemplars. Due to the complementarity from the categorylevel modeling, our method employs a lightweight architecture but achieves new high performance on three benchmarks. In addition, using a spatio-temporal network to tackle video frames, our method makes a good trade-off between effectiveness and efficiency. Code is available at https://github.com/VividLe/Online-Action-Detection.

Fine-grained sea-ice classification in SAR images is challenging due to the scarce labeled data and the imperfect annotation. Pre-training strategies are commonly carried out to prevent severe overfitting with limited labeled data. In spite of this, the observation noise still exists in the transferred features, which can be captured by aleatoric uncertainty. In this paper, we propose an aleatoric uncertainty embedded sea-ice classification method together with transfer learning of two different pre-training strategies. Instead of representing the transferred feature as a deterministic embedding, the proposed method concerns the feature uncertainty and models the embedding as a Gaussian distribution with variance. The experiments demonstrate that the proposed aleatoric uncertainty estimation is beneficial to improving the classification result of transfer learning. Based on the measured feature uncertainty, we analyze the potential of integrating two different pre-trained models to further enhance the performance.

Few-shot segmentation, which aims to segment unseen-class objects given only a handful of densely labeled samples, has received widespread attention from the community. Existing approaches typically follow the prototype learning paradigm to perform meta-inference, which fails to fully exploit the underlying information from support image-mask pairs, resulting in various segmentation failures, e.g., incomplete objects, ambiguous boundaries, and distractor activation. To this end, we propose a simple yet versatile framework in the spirit of divide-and-conquer. Specifically, a novel self-reasoning scheme is first implemented on the annotated support image, and then the coarse segmentation mask is divided into multiple regions with different properties. Leveraging effective masked average pooling operations, a series of support-induced proxies are thus derived, each playing a specific role in conquering the above challenges. Moreover, we devise a unique parallel decoder structure that integrates proxies with similar attributes to boost the discrimination power. Our proposed approach, named divide-and-conquer proxies (DCP), allows for the development of appropriate and reliable information as a guide at the “episode” level, not just about the object cues themselves. Extensive experiments on PASCAL-5i and COCO-20i demonstrate the superiority of DCP over conventional prototype-based approaches (up to 5~10% on average), which also establishes a new state-of-the-art. Code is available at github.com/chunbolang/DCP.

Improving the resolution of magnetic resonance (MR) image data is critical to computer-aided diagnosis and brain function analysis. Higher resolution helps to capture more detailed content, but typically induces to lower signal-to-noise ratio and longer scanning time. To this end, MR image super-resolution has become a widely-interested topic in recent times. Existing works establish extensive deep models with the conventional architectures based on convolutional neural networks (CNN). In this work, to further advance this research field, we make an early effort to build a Transformer-based MR image super-resolution framework, with careful designs on exploring valuable domain prior knowledge. Specifically, we consider two-fold domain priors including the high-frequency structure prior and the inter-modality context prior, and establish a novel Transformer architecture, called Cross-modality high-frequency Transformer (Cohf-T), to introduce such priors into super-resolving the low-resolution (LR) MR images. Experiments on two datasets indicate that Cohf-T achieves new state-of-the-art performance.

Few-shot semantic segmentation aims to segment the target objects in query under the condition of a few annotated support images. Most previous works strive to mine more effective category information from the support to match with the corresponding objects in query. However, they all ignored the category information gap between query and support images. If the objects in them show large intra-class diversity, forcibly migrating the category information from the support to the query is ineffective. To solve this problem, we are the first to introduce an intermediate prototype for mining both deterministic category information from the support and adaptive category knowledge from the query. Specifically, we design an Intermediate Prototype Mining Transformer (IPMT) to learn the prototype in an iterative way. In each IPMT layer, we propagate the object information in both support and query features to the prototype and then use it to activate the query feature map. By conducting this process iteratively, both the intermediate prototype and the query feature can be progressively improved. At last, the final query feature is used to yield precise segmentation prediction. Extensive experiments on both PASCAL-5i and COCO-20i datasets clearly verify the effectiveness of our IPMT and show that it outperforms previous state-of-the-art methods by a large margin. Code is available at https://github.com/LIUYUANWEI98/IPMT

This paper proposes an end-to-end fine-grained visual categorization system, termed Part-based Convolutional Neural Network (P-CNN), which consists of three modules. The first module is a Squeeze-and-Excitation (SE) block, which learns to recalibrate channel-wise feature responses by emphasizing informative channels and suppressing less useful ones. The second module is a Part Localization Network (PLN) used to locate distinctive object parts, through which a bank of convolutional filters are learned as discriminative part detectors. Thus, a group of informative parts can be discovered by convolving the feature maps with each part detector. The third module is a Part Classification Network (PCN) that has two streams. The first stream classifies each individual object part into image-level categories. The second stream concatenates part features and global feature into a joint feature for the final classification. In order to learn powerful part features and boost the joint feature capability, we propose a Duplex Focal Loss used for metric learning and part classification, which focuses on training hard examples. We further merge PLN and PCN into a unified network for an end-to-end training process via a simple training technique. Comprehensive experiments and comparisons with state-of-the-art methods on three benchmark datasets demonstrate the effectiveness of our proposed method.

Conventional salient object detection models cannot differentiate the importance of different salient objects. Recently, two works have been proposed to detect saliency ranking by assigning different degrees of saliency to different objects. However, one of these models cannot differentiate object instances and the other focuses more on sequential attention shift order inference. In this paper, we investigate a practical problem setting that requires simultaneously segment salient instances and infer their relative saliency rank order. We present a novel unified model as the first end-to-end solution, where an improved Mask R-CNN is first used to segment salient instances and a saliency ranking branch is then added to infer the relative saliency. For relative saliency ranking, we build a new graph reasoning module by combining four graphs to incorporate the instance interaction relation, local contrast, global contrast, and a high-level semantic prior, respectively. A novel loss function is also proposed to effectively train the saliency ranking branch. Besides, a new dataset and an evaluation metric are proposed for this task, aiming at pushing forward this field of research. Finally, experimental results demonstrate that our proposed model is more effective than previous methods. We also show an example of its practical usage on adaptive image retargeting.

How to effectively fuse cross-modal information is a key problem for RGB-D salient object detection. Early fusion and result fusion schemes fuse RGB and depth information at the input and output stages, respectively, and hence incur distribution gaps or information loss. Many models instead employ a feature fusion strategy, but they are limited by their use of low-order point-to-point fusion methods. In this paper, we propose a novel mutual attention model by fusing attention and context from different modalities. We use the non-local attention of one modality to propagate long-range contextual dependencies for the other, thus leveraging complementary attention cues to achieve high-order and trilinear cross-modal interaction. We also propose to induce contrast inference from the mutual attention and obtain a unified model. Considering that low-quality depth data may be detrimental to model performance, we further propose a selective attention to reweight the added depth cues. We embed the proposed modules in a two-stream CNN for RGB-D SOD. Experimental results demonstrate the effectiveness of our proposed model. Moreover, we also construct a new and challenging large-scale RGB-D SOD dataset of high-quality, which can promote both the training and evaluation of deep models.

In this paper, we reveal that metric learning would suffer from serious inseparable problem if without informative sample mining. Since the inseparable samples are often mixed with hard samples, current informative sample mining strategies used to deal with inseparable problem may bring up some side-effects, such as instability of objective function, etc. To alleviate this problem, we propose a novel distance metric learning algorithm, named adaptive neighborhood metric learning (ANML). In ANML, we design two thresholds to adaptively identify the inseparable similar and dissimilar samples in the training procedure, thus inseparable sample removing and metric parameter learning are implemented in the same procedure. Due to the non-continuity of the proposed ANML, we develop an ingenious function, named log-exp mean function to construct a continuous formulation to surrogate it, which can be efficiently solved by the gradient descent method. Similar to Triplet loss, ANML can be used to learn both the linear and deep embeddings. By analyzing the proposed method, we find it has some interesting properties. For example, when ANML is used to learn the linear embedding, current famous metric learning algorithms such as the large margin nearest neighbor (LMNN) and neighbourhood components analysis (NCA) are the special cases of the proposed ANML by setting the parameters different values. When it is used to learn deep features, the state-of-the-art deep metric learning algorithms such as Triplet loss, Lifted structure loss, and Multi-similarity loss become the special cases of ANML. Furthermore, the log-exp mean function proposed in our method gives a new perspective to review the deep metric learning methods such as Prox-NCA and N-pairs loss. At last, promising experimental results demonstrate the effectiveness of the proposed method.

Weakly supervised temporal action localization aims at learning the instance-level action pattern from the video-level labels, where a significant challenge is action-context confusion. To overcome this challenge, one recent work builds an action-click supervision framework. It requires similar annotation costs but can steadily improve the localization performance when compared to the conventional weakly supervised methods. In this paper, by revealing that the performance bottleneck of the existing approaches mainly comes from the background errors, we find that a stronger action localizer can be trained with labels on the background video frames rather than those on the action frames. To this end, we convert the action-click supervision to the background-click supervision and develop a novel method, called BackTAL. Specifically, BackTAL implements two-fold modeling on the background video frames, i.e., the position modeling and the feature modeling. In position modeling, we not only conduct supervised learning on the annotated video frames but also design a score separation module to enlarge the score differences between the potential action frames and backgrounds. In feature modeling, we propose an affinity module to measure frame-specific similarities among neighboring frames and dynamically attend to informative neighbors when calculating temporal convolution. Extensive experiments on three benchmarks are conducted, which demonstrate the high performance of the established BackTAL and the rationality of the proposed background-click supervision.

As an emerging and challenging problem in the computer vision community, weakly supervised object localization and detection plays an important role for developing new generation computer vision systems and has received significant attention in the past decade. As methods have been proposed, a comprehensive survey of these topics is of great importance. In this work, we review (1) classic models, (2) approaches with feature representations from off-the-shelf deep networks, (3) approaches solely based on deep learning, and (4) publicly available datasets and standard evaluation metrics that are widely used in this field. We also discuss the key challenges in this field, development history of this field, advantages/disadvantages of the methods in each category, the relationships between methods in different categories, applications of the weakly supervised object localization and detection methods, and potential future directions to further promote the development of this research field.

In recent years, weakly supervised object detection has attracted great attention in the computer vision community. Although numerous deep learning-based approaches have been proposed in the past few years, such an ill-posed problem is still challenging and the learning performance is still behind the expectation. In fact, most of the existing approaches only consider the visual appearance of each proposal region but ignore to make use of the helpful context information. To this end, this paper introduces two levels of context into the weakly supervised learning framework. The first one is the proposal-level context, i.e., the relationship of the spatially adjacent proposals. The second one is the semantic-level context, i.e., the relationship of the co-occurring object categories. Therefore, the proposed weakly supervised learning framework contains not only the cognition process on the visual appearance but also the reasoning process on the proposal- and semantic-level relationships, which leads to the novel deep multiple instance reasoning framework. Specifically, built upon a conventional CNN-based network architecture, the proposed framework is equipped with two additional graph convolutional network-based reasoning models to implement object location reasoning and multi-label reasoning within an end-to-end network training procedure. Comprehensive experiments on the widely used PASCAL VOC and MS COCO benchmarks have been implemented, which demonstrate the superior capacity of the proposed approach when compared with other state-of-the-art methods and baseline models.

Video object segmentation and detection are two important tasks toward intelligent video content understanding. Due to their wide applications in real-world vision tasks, such as video surveillance and automatic driving, they have recently attracted great attention in the computer vision and multimedia processing communities. Although numerous deep learning-based approaches have been proposed to solve these problems, implementing effective and efficient video object segmentation and detection is still very challenging for now, and the principles of solutions to address the problems are still understudied. On the one hand, the features learned by the current deep models are not strong enough to capture the rich spatial and temporal information from the input videos. On the other hand, the annotation information in the video data (especially for unconstrained online videos) is usually insufficient, unspecific, or even absent, thus challenging the current mainstream learning schemes.

Few-shot image classification has attracted extensive attention, which aims to recognize unseen classes given only a few labeled samples. Due to the large intraclass variances and interclass similarity of remote sensing scenes, the task under such circumstance is much more challenging than general few-shot image classification. Most existing prototype-based few-shot algorithms usually calculate prototypes directly from support samples and ignore the validity of prototypes, which results in a decline in the accuracy of subsequent inferences based on prototypes. To tackle this problem, we propose a Siamese-prototype network (SPNet) with prototype self-calibration (SC) and intercalibration (IC). First, to acquire more accurate prototypes, we utilize the supervision information from support labels to calibrate the prototypes generated from support features. This process is called SC. Second, we propose to consider the confidence scores of the query samples as another type of prototypes, which are then used to predict the support samples in the same way. Thus, the information interaction between support and query samples is implicitly a further calibration for prototypes (so-called IC). Our model is optimized with three losses, of which two additional losses help the model to learn more representative prototypes and make more accurate predictions. With no additional parameters to be learned, our model is very lightweight and convenient to employ. The experiments on three public remote sensing image datasets demonstrate competitive performance compared with other advanced few-shot image classification approaches. The source code is available athttps://github.com/zoraup/SPNet.

The methods for remote sensing image (RSI) scene classification based on deep convolutional neural networks (DCNNs) have achieved prominent success. However, confronted with adversarial examples obtained by adding imperceptible perturbations to clean images, the great vulnerability of DCNNs makes it worth exploring effective defense methods. To date, numerous countermeasures for adversarial examples have been proposed, but how to improve the defensive ability for unknown attacks still to be answered. To address this issue, in this article, we propose an effective defense framework specified for RSI scene classification, named perturbation-seeking generative adversarial networks (PSGANs). In brief, a new training framework is designed to train the classifier by introducing the examples generated during the image reconstruction process, in addition to clean examples and adversarial ones. These generated examples can be random kinds of unknown attacks during training and thus are utilized to eliminate the blind spots of a classifier. To assist the proposed training framework, a reconstruction method is developed. First, instead of modeling the distribution of clean examples, we model the distributions of the perturbations added in adversarial examples. Second, to make a tradeoff between the diversity of the reconstructed examples and the optimization of PSGAN, a scale factor named seeking radius is introduced to scale the generated perturbations before they are subtracted by the given adversarial examples. Comprehensive and extensive experimental results on three widely used benchmarks for RSI scene classification demonstrate the great effectiveness of PSGAN when faced with both known and unknown attacks. Our source code is available athttps://github.com/xuxiangsun/PSGAN.

Deep learning has substantially pushed forward remote sensing image change detection through extracting discriminative hierarchical features. However, as the increasingly high resolution remote sensing images have abundant spatial details but limited spectral information, the use of conventional backbone networks would give rise to blurry boundaries between different semantics among hierarchical features. This explains why most false alarms in the final predictions distribute around change boundaries. To alleviate the problem, we pay attention to feature refinement and propose deep learning networks that deliver improved separability (ISNet). Our ISNet reaps the advantages from two strategies applied to refining bi-temporal feature hierarchies: (i) margin maximization that clarifies the gap between changed and unchanged semantics, and (ii) targeted arrangement of attention mechanisms that directs the use of channel attention and spatial attention for highlighting semantic and positional information, respectively. Specifically, we insert channel attention modules into share-weighted backbone networks to facilitate semantic-specific feature extraction. The semantic boundaries in the extracted bi-temporal hierarchical features are then clarified by margin maximization modules, followed by spatial attention modules to enhance positional change responses. A top-down fusion pathway makes the final refined features cover multi-scale representations and have strong separability for remote sensing image change detection. Extensive experimental evaluations demonstrate that our ISNet achieves state-of-the-art performance on the LEVIR-CD, SYSU-CD, and Season-Varying datasets, in terms of Overall Accuracy (OA), Intersection-of-Union (IoU), and F1 score. Code is available at https://github.com/xingronaldo/ISNet.

Oriented object detection is a practical and challenging task in remote sensing image interpretation. Nowadays, oriented detectors mostly use horizontal boxes as intermedium to derive oriented boxes from them. However, the horizontal boxes are inclined to get small Intersection-over-Unions (IoUs) with ground truths, which may have some undesirable effects, such as introducing redundant noise, mismatching with ground truths, detracting from the robustness of detectors, etc. In this paper, we propose a novel Anchor-free Oriented Proposal Generator (AOPG) that abandons horizontal box-related operations from the network architecture. AOPG first produces coarse oriented boxes by a Coarse Location Module (CLM) in an anchor-free manner and then refines them into high-quality oriented proposals. After AOPG, we apply a Fast R-CNN head to produce the final detection results. Furthermore, the shortage of large-scale datasets is also a hindrance to the development of oriented object detection. To alleviate the data insufficiency, we release a new dataset on the basis of our DIOR dataset and name it DIOR-R. Massive experiments demonstrate the effectiveness of AOPG. Particularly, without bells and whistles, we achieve the accuracy of 64.41%, 75.24% and 96.22% mAP on the DIOR-R, DOTA and HRSC2016 datasets respectively. Code and models are available at https://github.com/jbwang1997/AOPG.

Weakly Supervised Object Detection (WSOD) in remote sensing images remains a challenging task when learning object detectors with only image-level labels. As we know, object proposal generation plays a crucial role in WSOD. At present, the proposal generation of most existing WSOD methods mainly relies on heuristic strategies such as selective search and Edge Boxes. However, the proposals obtained by the above methods cannot well cover the entire objects, severely hindering the performance of WSOD. To address this issue, this paper proposes a Self-guided Proposal Generation approach, termed SPG. It can be easily implemented with most WSOD methods in a unified framework. To this end, we first introduce a confidence propagation approach to obtain the objectness confidence map for each image, which on the one hand highlights informative object locations, and on the other hand aggregates discriminative feature representation by combining the objectness confidence map with the deep features. Then, the proposal generation is implemented by mining informative regions as proposals on the objectness confidence map. Extensive evaluations on two challenging datasets demonstrate that our SPG significantly improves the baseline methods Online Instance Classifier Refinement (OICR) and Min-Entropy Latent Model (MELM) by large margins (for OICR: 15.86% mAP and 12.89% CorLoc gains on the NWPU VHR-10.v2 dataset, 3.65% mAP and 4.87% CorLoc gains on the DIOR dataset; for MELM: 20.51% mAP and 23.54% CorLoc gains on the NWPU VHR-10.v2 dataset, 7.11% mAP and 4.96% CorLoc gains on the DIOR dataset) and achieves the state-of-the-art results compared with existing methods.

In the past few years, object detection in remote sensing images has achieved remarkable progress. However, the detection of oriented and densely packed objects are still unsatisfactory due to the following spatial and feature misalignments. 1) Most two-stage oriented detectors only introduce an orientation regression branch in the detection head, while still leverage horizontal proposals for classification and regression. This inevitably results in the spatial misalignment problem between horizontal proposals and oriented objects. 2) The features used for classification are in fact extracted from the region proposals which have shifted to the final predictions via the regression branch. This leads to the feature misalignment problem between the classification and the localization tasks. In this article, we present a two-stage oriented object detection method, termed dual-aligned oriented detector (DODet), toward evading the aforementioned problems of spatial and feature misalignments. In DODet, the first stage is an oriented proposal network (OPN), which generates high-quality oriented proposals via a novel representation scheme of oriented objects. The second stage is a localization-guided detection head (LDH) that aims at alleviating the feature misalignment between classification and localization. Comprehensive and extensive evaluations on three benchmarks, including DIOR-R, DOTA, and HRSC2016, indicate that our method could obtain consistent and substantial gains compared with the baseline method. The source code is publicly available athttps://github.com/yanqingyao1994/DODet.

Recently, due to the excellent representation ability of convolutional neural networks (CNNs), object detection in remote sensing images has undergone remarkable development. However, when trained with a small number of samples, the performance of the object detectors drops sharply. In this article, we focus on the following three main challenges of few-shot object detection in remote sensing images: 1) since the sample number of novel classes is far less than base classes, object detectors would fail to quickly adapt to the features of novel classes, which would result in overfitting; 2) the scarcity of samples in novel classes leads to a sparse orientation space, while the objects in remote sensing images usually have arbitrary orientations; and 3) the distribution of object instances in remote sensing images is scattered and, therefore, it is hard to identify foreground objects from the complex background. To tackle these problems, we propose a simple yet effective method named prototype-CNN (P-CNN), which mainly consists of three parts: a prototype learning network (PLN) converting support images to class-aware prototypes, a prototype-guided region proposal network (P-G RPN) for better generation of region proposals, and a detector head extending the head of Faster region-based CNN (R-CNN) to further boost the performance. Comprehensive evaluations on the large-scale DIOR dataset demonstrate the effectiveness of our P-CNN. The source code is available athttps://github.com/Ybowei/P-CNN.

Weakly supervised object detection (WSOD) in remote sensing images (RSIs) remains a challenge when learning a subtle object detection model with only image-level annotations. Most works tend to optimize the detection model via exploiting the most contributed region, thereby to be dominated by the most discriminative part of an object. Meanwhile, these methods ignore the consistency across different spatial transformations of the same image and always label them with different classes, which introduces potential ambiguities. To tackle these challenges, we propose a unique self-supervised adversarial and equivariant network (SAENet) and aim at learning complementary and consistent visual patterns for WSOD in RSIs. To this end, an adversarial dropout–activation block is first designed to facilitate the entire object detector via adaptively hiding the discriminative parts and highlighting the instance-related regions. Besides, we further introduce a flexible self-supervised transformation equivariance mechanism on each potential instance from multiple spatial transformations to obtain spatially consistent self-supervisions. Accordingly, the obtained supervisions can be leveraged to pursue a more robust and spatially consistent object detector. Comprehensive experiments on the challenging LEarning, VIsion and Remote sensing Laboratory (LEVIR), NorthWestern Polytechnical University (NWPU) VHR-10.v2, and detection in optical RSIs (DIOR) datasets validate that SAENet outperforms the previous state-of-the-art works and achieves 46.2%, 60.7%, and 27.1% mAP, respectively.

Few-shot learning (FSL), which aims to rapidly recognize unseen categories with limited samples, has attracted wide attention in aerial image scene classification. However, the existing methods generally train and evaluate the model within a dataset, and changing the dataset requires retraining and evaluation, which only realizes the generalization of intra-dataset. Considering meta-learning, this brings in a natural assumption: FSL should learn meta-knowledge from cross-domain heterogeneous tasks and then can generalize to new data distributions (e.g., datasets) with few samples. To this end, we propose a new benchmark, dubbed aerial image few-shot dataset (AIFS-DATASET), which is composed of diverse datasets and can provide more realistic heterogeneous task distributions. On AIFS-DATASET, we use many heterogeneous tasks, across multi-domains without any aerial image category, to train the model, achieving “see more.” Then we transfer the learned knowledge to new tasks in aerial images to evaluate the generalization performance of the model, thus acquiring a “well-informed” few-shot aerial image scene classification model. Moreover, the challenges of inter-class similarity and intra-class discrepancy in aerial images still exist. We also develop a dual constrained distance metric learning (DC-DML) framework to deal with the variable learning tasks adaptively and to achieve compact data distribution within a class and clear distribution gaps between classes from the perspective of metric learning. DC-DML mainly uses a task-adapted feature extractor while devising a novel distance metric with a cross-class bias penalty. By conducting experiments on AIFS-DATASET, we observed that DC-DML outperforms the current prevailing FSL approaches by a large margin.

In this article, we attempt to achieve one-shot object detection by mimicking the human ability to learn new concepts under limited reference, which aims at detecting all object instances of an unseen class in a target image when given a query image of the same unseen class. However, this one-shot learning ability of human benefits from the fact that human brain can quickly extract and process the associated information between the query–target images, which is an issue for the one-shot object detection framework to overcome. Moreover, the feature extraction of the query class in target images is intractable due to the complex and diversified background of remote sensing images. To solve these issues, we propose a solo-to-collaborative dual-attention network (SCoDANet) to hierarchically (image itself/pairs) enhance image feature representations. It consists of three components: 1) solo-attention head that strengthens the compactness of intraclass feature representations of an image and avoids background interference by selectively aggregating the similar features from the spatial and channel dimensions, respectively; 2) dual coattention module that guides RPN to generate an expected set of region proposals related to the query class by mining the coinformation of each query–target feature pair; and 3) nonlinear matching that provides a measure of similarity between the query feature and proposals of the target image to further learn a more robust detector. Our extensive experiments over two benchmarks demonstrate the effectiveness of our method under the one-shot scenario of detecting seen and unseen object categories.

Few-shot semantic segmentation, aiming to segment query images with a few annotated support samples, has drawn increasing attention. Most existing few-shot methods leverage the single prototype obtained from global average pooling to represent all support information and further use the extracted prototype to segment the query images in a matching manner. Although promising results for natural images have been reported, these methods cannot be directly applied on aerial images. The main reason comes from that the extracted single support prototype can only provide a coarse guidance for matching between query and support images and could not handle the large variance of objects’ appearances and scales. To deal with these challenges on aerial images, we propose a scale-aware few-shot semantic segmentation network to perform detailed matching with multiple prototypes. More specifically, the detailed matching module is first constructed to compute the pixel-level similarity between the query features and the extracted multiple support prototypes for providing more accurate parsing guidance. Subsequently, to address the problem of scale imbalance, the scale-aware focal loss is designed to dynamically down-weight the loss assigned to large well-parsed objects and focus training on tiny hard-parsed objects. To facilitate the reproducible research on the task of few-shot semantic segmentation in aerial images, we further provide a few-shot segmentation benchmark iSAID-$5^{\mathrm {i}}$constructed from the large-scale iSAID dataset[1]. Comprehensive experiments and comparisons with the state-of-the-art few-shot segmentation methods on the iSAID-$5^{\mathrm {i}}$dataset clearly demonstrate the superiority of our proposed method. The code and dataset are available athttps://github.com/caoql98/SDM.

Anchor-free aerial object detection methods have recently attracted much attention due to their simplicity and efficiency. However, the performance is still unsatisfactory due to the following two main limitations. On the one hand, the anchor-free detector employs ordinary convolution layers with axis-aligned receptive fields to extract object features, resulting in lacking internal mechanisms to handle the rotation variance. On the other hand, the detector sacrifices much semantic information to achieve faster detection, leading to the inability to deal with objects’ high inter-class similarity and intra-class diversity. To address these issues, in this paper, we present a unique anchor-free detector, termed Rotation-Insensitive Point Representation (R2IPoints), of which a set of category-aware points are employed to encode the spatial and semantic information of the arbitrary-oriented objects. Specifically, we first devise a Stacked Rotation convolution Module (SRM) to encourage the learning of rotation-insensitive point representation by adaptively modelling orientation-agnostic interdependencies over stochastically rotated features. Meanwhile, we further introduce a Class-specific Semantic enhancement Module (CSM). It performs category-aware semantic activation to recalibrate features, thus enabling the point representation to be aware of object categories. Through jointly optimizing the two proposed modules in an end-to-end manner, R2IPoints could simultaneously generate rotation-insensitive and category-aware point representation. Extensive experiments on the challenging DIOR and DOTA datasets demonstrate the superiority of the proposed method. We achieve 72.7% mAP on DIOR and 74.34% mAP on DOTA, surpassing the baseline method of +2.4% mAP and +2.49% mAP, respectively. The code is available at https://github.com/shnew/R2IPoints.

Domain adaptation scene classification refers to the task of scene classification where the training set (called source domain) has different distributions from the test set (called target domain). Although remarkable results have been reported, the misalignment of source and target domain features still remains a big challenge when the large intraclass variances of remote sensing images encounter the insufficient exploration of discriminative feature representations for both domains. To address this challenge, a novel domain feature enhancement network (DFENet) is proposed to adaptively enhance the discriminative ability of the learned features for dealing with the domain variances of scene classification. Specifically, an adaptive context-aware feature refinement (CAFR) module is first designed to automatically recalibrate global and local features by explicitly modeling interdependencies between the channel and spatial for each domain. Then, a multilevel adversarial dropout (MAD) module is further designed to strengthen the generalization capability of our network by adaptively reconfiguring the sparsity of the feature level and decision level in the target domain. The cooperation of CAFR module and MAD module formulates a unique DFENet that can be learned in an end-to-end manner. Comprehensive experiments show that our proposed method is better than state-of-the-art methods on Merced$\to $RSSCN7, AID$\to $RSSCN7, NWPU$\to $RSSCN7, RSSCN7$\to $Merced, RSSCN7$\to $AID, and RSSCN7$\to $NWPU datasets.

RGB-D co-salient object detection aims to segment co-occurring salient objects when given a group of relevant images and depth maps. Previous methods often adopt separate pipeline and use hand-crafted features, being hard to capture the patterns of co-occurring salient objects and leading to unsatisfactory results. Using end-to-end CNN models is a straightforward idea, but they are less effective in exploiting global cues due to the intrinsic limitation. Thus, in this paper, we alternatively propose an end-to-end transformer-based model which uses class tokens to explicitly capture implicit class knowledge to perform RGB-D co-salient object detection, denoted as CTNet. Specifically, we first design adaptive class tokens for individual images to explore intra-saliency cues and then develop common class tokens for the whole group to explore inter-saliency cues. Besides, we also leverage the complementary cues between RGB images and depth maps to promote the learning of the above two types of class tokens. In addition, to promote model evaluation, we construct a challenging and large-scale benchmark dataset, named RGBD CoSal1k, which collects 106 groups containing 1000 pairs of RGB-D images with complex scenarios and diverse appearances. Experimental results on three benchmark datasets demonstrate the effectiveness of our proposed method.

Recently, many real-world applications exploit multi-view data, which is collected from diverse domains or obtained from various feature extractors and reflect different properties or distributions of the data. In this work, a novel unsupervised multi-view framework is proposed to cluster such data. The proposed method, called Multi-View clustering with Adaptive Sparse Memberships and Weight Allocation (MVASM), pays more attention to constructing a common membership matrix with proper sparseness over different views and learns the centroid matrix and its corresponding weight of each view. Concretely, MVASM method attempts to learn a common and flexible sparse membership matrix to indicate the clustering, which explores the underlying consensus information of multiple views, and solves the multiple centroid matrices and weights to utilize the view-specific information and further modifies the above-mentioned membership matrix. In addition, the theoretical analysis, including the determination of the power exponent parameter, convergence analysis, and complexity analysis are also presented. Compared to the state-of-the-art methods, the proposed method improves the performance of clustering on different public datasets and demonstrates its reasonability and superiority.

Weakly supervised object localization (WSOL) aims at learning to localize objects of interest by only using the image-level labels as the supervision. While numerous efforts have been made in this field, recent approaches still suffer from two challenges: one is the part domination issue while the other is the learning robustness issue. Specifically, the former makes the localizer prone to the local discriminative object regions rather than the desired whole object, and the latter makes the localizer over-sensitive to the variations of the input images so that one can hardly obtain localization results robust to the arbitrary visual stimulus. To solve these issues, we propose a novel framework to strengthen the learning tolerance, referred to as SLT-Net, for WSOL. Specifically, we consider two-fold learning tolerance strengthening mechanisms. One is the semantic tolerance strengthening mechanism, which allows the localizer to make mistakes for classifying similar semantics so that it will not concentrate too much on the discriminative local regions. The other is the visual stimuli tolerance strengthening mechanism, which enforces the localizer to be robust to different image transformations so that the prediction quality will not be sensitive to each specific input image. Finally, we implement comprehensive experimental comparisons on two widely-used datasets CUB and ILSVRC2012, which demonstrate the effectiveness of our proposed approach.

Significant performance improvement has been achieved for fully-supervised video salient object detection with the pixel-wise labeled training datasets, which are time-consuming and expensive to obtain. To relieve the burden of data annotation, we present the first weakly super-vised video salient object detection model based on relabeled “fixation guided scribble annotations”. Specifically, an "Appearance-motion fusion module" and bidirectional ConvLSTM based framework are proposed to achieve effective multi-modal learning and long-term temporal context modeling based on our new weak annotations. Further, we design a novel foreground-background similarity loss to further explore the labeling similarity across frames. A weak annotation boosting strategy is also introduced to boost our model performance with a new pseudo-label generation technique. Extensive experimental results on six benchmark video saliency detection datasets illustrate the effectiveness of our solution1.

Deep learning (DL)-based methods have achieved great success in solving the ill-posed JPEG compression artifacts removal problem. However, as most DL architectures are designed to directly learn pixel-level mapping relationship-s, they largely ignore semantic-level information and lack sufficient interpretability. To address the above issues, in this work, we propose an interpretable deep network to learn both pixel-level regressive prior and semantic-level discriminative prior. Specifically, we design a variational model to formulate the image de-blocking problem and propose two prior terms for the image content and gradient, respectively. The content-relevant prior is formulated as a DL-based image-to-image regressor to perform as a de-blocker from the pixel-level. The gradient-relevant prior serves as a DL-based classifier to distinguish whether the image is compressed from the semantic-level. To effectively solve the variational model, we design an alternating minimization algorithm and unfold it into a deep network architecture. In this way, not only the interpretability of the deep network is increased, but also the dual priors can be well estimated from training samples. By integrating the two priors into a single framework, the image de-blocking problem can be well-constrained, leading to a better performance. Experiments on benchmarks and real-world use cases demonstrate the superiority of our method to the existing state-of-the-art approaches.

Existing state-of-the-art saliency detection methods heavily rely on CNN-based architectures. Alternatively, we rethink this task from a convolution-free sequence-to-sequence perspective and predict saliency by modeling long-range dependencies, which can not be achieved by convolution. Specifically, we develop a novel unified model based on a pure transformer, namely, Visual Saliency Transformer (VST), for both RGB and RGB-D salient object detection (SOD). It takes image patches as inputs and leverages the transformer to propagate global contexts among image patches. Unlike conventional architectures used in Vision Transformer (ViT), we leverage multi-level token fusion and propose a new token upsampling method under the transformer framework to get high-resolution detection results. We also develop a token-based multi-task decoder to simultaneously perform saliency and boundary detection by introducing task-related tokens and a novel patch-task-attention mechanism. Experimental results show that our model outperforms existing methods on both RGB and RGB-D SOD benchmark datasets. Most importantly, our whole framework not only provides a new perspective for the SOD field but also shows a new paradigm for transformer-based dense prediction models. Code is available at https://github.com/nnizhang/VST.

The application of light field data in salient object detection is becoming increasingly popular recently. The difficulty lies in how to effectively fuse the features within the focal stack and how to cooperate them with the feature of the all-focus image. Previous methods usually fuse focal stack features via convolution or ConvLSTM, which are both less effective and ill-posed. In this paper, we model the information fusion within focal stack via graph networks. They introduce powerful context propagation from neighbouring nodes and also avoid ill-posed implementations. On the one hand, we construct local graph connections thus avoiding prohibitive computational costs of traditional graph networks. On the other hand, instead of processing the two kinds of data separately, we build a novel dual graph model to guide the focal stack fusion process using all-focus patterns. To handle the second difficulty, previous methods usually implement one-shot fusion for focal stack and all-focus features, hence lacking a thorough exploration of their supplements. We introduce a reciprocative guidance scheme and enable mutual guidance between these two kinds of information at multiple steps. As such, both kinds of features can be enhanced iteratively, finally benefiting the saliency prediction. Extensive experimental results show that the proposed models are all beneficial and we achieve significantly better results than state-of-the-art methods.

Current state-of-the-art two-stage detectors generate oriented proposals through time-consuming schemes. This diminishes the detectors’ speed, thereby becoming the computational bottleneck in advanced oriented object detection systems. This work proposes an effective and simple oriented object detection framework, termed Oriented R-CNN, which is a general two-stage oriented detector with promising accuracy and efficiency. To be specific, in the first stage, we propose an oriented Region Proposal Network (oriented RPN) that directly generates high-quality oriented proposals in a nearly cost-free manner. The second stage is oriented R-CNN head for refining oriented Regions of Interest (oriented RoIs) and recognizing them. Without tricks, oriented R-CNN with ResNet50 achieves state-of-the-art detection accuracy on two commonly-used datasets for oriented object detection including DOTA (75.87% mAP) and HRSC2016 (96.50% mAP), while having a speed of 15.1 FPS with the image size of 1024×1024 on a single RTX 2080Ti. We hope our work could inspire rethinking the design of oriented detectors and serve as a baseline for oriented object detection. Code is available at https://github.com/jbwang1997/OBBDetection.

Humans perform co-saliency detection by first summarizing the consensus knowledge in the whole group and then searching corresponding objects in each image. Previous methods usually lack robustness, scalability, or stability for the first process and simply fuse consensus features with image features for the second process. In this paper, we propose a novel consensus-aware dynamic convolution model to explicitly and effectively perform the "summarize and search" process. To summarize consensus image features, we first summarize robust features for every single image using an effective pooling method and then aggregate cross-image consensus cues via the self-attention mechanism. By doing this, our model meets the scalability and stability requirements. Next, we generate dynamic kernels from consensus features to encode the summarized consensus knowledge. Two kinds of kernels are generated in a supplementary way to summarize fine-grained image-specific consensus object cues and the coarse group-wise common knowledge, respectively. Then, we can effectively perform object searching by employing dynamic convolution at multiple scales. Besides, a novel and effective data synthesis method is also proposed to train our network. Experimental results on four benchmark datasets verify the effectiveness of our proposed method. Our code and saliency maps are available at https://github.com/nnizhang/CADC.

Category-specific 3D object shape models have greatly boosted the recent advances in object detection, recognition and segmentation. However, even the most advanced approach for learning 3D object shapes still requires heavy manual annotations on large-scale 2D images. Such annotations include object categories, object keypoints, and figure-ground segmentation for the instances in each image. In particular, annotating figure-ground segmentation is unbearably labor-intensive and time-consuming. To address this problem, this paper devotes to learn category-specific 3D shape models under weak supervision, where only object categories and keypoints are required to be manually annotated on the training 2D images. By exploring the underlying relationship between two tasks: object segmentation and category-specific 3D shape reconstruction, we propose a novel weakly-supervised learning framework to jointly address these two tasks and combine them to boost the final performance of the learned 3D shape models. Moreover, learning without using figure-ground segmentation leads to ambiguous solutions. To this end, we develop the confidence weighting schemes in the viewpoint estimation and 3D shape learning procedure. These schemes effectively reduce the confusion caused by the noisy data and thus increase the chances for recovering more reliable 3D object shapes. Comprehensive experiments on the challenging PASCAL VOC benchmark show that our framework achieves comparable performance with the state-of-the-art methods that use expensive manual segmentation-level annotations. In addition, our experiments also demonstrate that our 3D shape models improve object segmentation performance.

In recent years, predicting the saccadic scanpaths of humans has become a new trend in the field of visual attention modeling. Given various saccadic algorithms, determining how to evaluate their ability to model a dynamic saccade has become an important yet understudied issue. To our best knowledge, existing metrics for evaluating saccadic prediction models are often heuristically designed, which may produce results that are inconsistent with human subjective assessment. To this end, we first construct a subjective database by collecting the assessments on 5,000 pairs of scanpaths from ten subjects. Based on this database, we can compare different metrics according to their consistency with human visual perception. In addition, we also propose a data-driven metric to measure scanpath similarity based on the human subjective comparison. To achieve this goal, we employ a long short-term memory (LSTM) network to learn the inference from the relationship of encoded scanpaths to a binary measurement. Experimental results have demonstrated that the LSTM-based metric outperforms other existing metrics. Moreover, we believe the constructed database can be used as a benchmark to inspire more insights for future metric selection.

Brain imaging genetics studies the genetic basis of brain structures and functionalities via integrating genotypic data such as single nucleotide polymorphisms (SNPs) and imaging quantitative traits (QTs). In this area, both multi-task learning (MTL) and sparse canonical correlation analysis (SCCA) methods are widely used since they are superior to those independent and pairwise univariate analysis. MTL methods generally incorporate a few of QTs and could not select features from multiple QTs; while SCCA methods typically employ one modality of QTs to study its association with SNPs. Both MTL and SCCA are computational expensive as the number of SNPs increases. In this paper, we propose a novel multi-task SCCA (MTSCCA) method to identify bi-multivariate associations between SNPs and multi-modal imaging QTs. MTSCCA could make use of the complementary information carried by different imaging modalities. MTSCCA enforces sparsity at the group level via the${\mathrm G}_{2,1}$-norm, and jointly selects features across multiple tasks for SNPs and QTs via the$\ell _{2,1}$-norm. A fast optimization algorithm is proposed using the grouping information of SNPs. Compared with conventional SCCA methods, MTSCCA obtains better correlation coefficients and canonical weights patterns. In addition, MTSCCA runs very fast and easy-to-implement, indicating its potential power in genome-wide brain-wide imaging genetics.

Weakly supervised object detection (WSOD) in remote sensing images (RSI) plays an essential role in RSI understanding applications. Currently, predominant works are inclined to first activate the most discriminative region and then pursue the whole object by analyzing the context information of the activated region. However, the most discriminative region usually only covers a small crucial part. Besides, many same-class instances often appear in adjacent locations. In such a case, treating proposals of large spatial overlap as the same-class instances not only introduces potential ambiguities but also misleads the detection model to recognize multiple adjacent instances as one object instance. To address these challenges, a novel triple context-aware network (TCANet) is proposed to learn complementary and discriminative visual patterns for WSOD in RSIs. Specifically, a global context-aware enhancement (GCAE) module is first designed to activate the features of the whole object by capturing the global visual scene context. Then, a dual-local context residual (DLCR) module is further developed to capture the instance-level discriminative cues by leveraging the semantic discrepancy of the local context. Furthermore, an effective adaptive-weighted refinement loss is integrated into the DLCR module to reduce the ambiguities in the label propagating process. The collaboration of GCAE and DLCR formulates a unique TCANet that can be learned in an end-to-end manner. Comprehensive experiments are carried out on the challenging NWPU VHR-10.v2 and DIOR data sets. We achieve a 58.8% mAP and a 25.8% mAP on the NWPU VHR-10.v2 and DIOR data sets, respectively, which both significantly outperform the state of the arts.

Few-shot scene classification aims to recognize unseen scene concepts from few labeled samples. However, most existing works are generally inclined to learn metalearners or transfer knowledge while ignoring the importance to learn discriminative representations and a proper metric for remote sensing images. To address these challenges, in this article, we propose an end-to-end network for boosting a few-shot remote sensing image scene classification, called discriminative learning of adaptive match network (DLA-MatchNet). Specifically, we first adopt the attention technique to delve into the interchannel and interspatial relationships to automatically discover discriminative regions. Then, the channel attention and spatial attention modules can be incorporated with the feature network by using different feature fusion schemes, achieving “discriminative learning.” Afterward, considering the issues of the large intraclass variances and interclass similarity of remote sensing images, instead of simply computing the distances between the support samples and query samples, we concatenate the support and query discriminative features in depth and utilize a matcher to “adaptively” select the semantically relevant sample pairs to assign similarity scores. Our method leverages an episode-based strategy to train the model. Once trained, our model can predict the category of query image without further fine-tuning. Experimental results on three public remote sensing image data sets demonstrate the effectiveness of our model in the few-shot scene classification task.

In this article, we focus on tackling the problem of weakly supervised object detection from high spatial resolution remote sensing images, which aims to learn detectors with only image-level annotations, i.e., without object location information during the training stage. Although promising results have been achieved, most approaches often fail to provide high-quality initial samples and thus are difficult to obtain optimal object detectors. To address this challenge, a dynamic curriculum learning strategy is proposed to progressively learn the object detectors by feeding training images with increasing difficulty that matches current detection ability. To this end, an entropy-based criterion is firstly designed to evaluate the difficulty for localizing objects in images. Then, an initial curriculum that ranks training images in ascending order of difficulty is generated, in which easy images are selected to provide reliable instances for learning object detectors. With the gained stronger detection ability, the subsequent order in the curriculum for retraining detectors is accordingly adjusted by promoting difficult images as easy ones. In such way, the detectors can be well prepared by training on easy images for learning from more difficult ones and thus gradually improve their detection ability more effectively. Moreover, an effective instance-aware focal loss function for detector learning is developed to alleviate the influence of positive instances of bad quality and meanwhile enhance the discriminative information of class-specific hard negative instances. Comprehensive experiments and comparisons with state-of-the-art methods on two publicly available data sets demonstrate the superiority of our proposed method.

Different from conventional instance segmentation, salient instance segmentation (SIS) faces two difficulties. The first is that it involves segmenting salient instances only while ignoring background, and the second is that it targets generic object instances without pre-defined object categories. In this paper, based on the state-of-the-art Mask R-CNN model, we propose to leverage complementary saliency and contour information to handle these two challenges. We first improve Mask R-CNN by introducing an interleaved execution strategy and proposing a novel mask head network to incorporate global context within each RoI. Then we add two branches to Mask R-CNN for saliency and contour detection, respectively. We fuse the Mask R-CNN features with the saliency and contour features, where the former supply pixel-wise saliency information to help with identifying salient regions and the latter provide a generic object contour prior to help detect and segment generic objects. We also propose a novel multiscale global attention model to generate attentive global features from multiscale representative features for feature fusion. Experimental results demonstrate that all our proposed model components can improve SIS performance. Finally, our overall model outperforms state-of-the-art SIS methods and Mask R-CNN by more than 6% and 3%, respectively. By using additional multitask training data, we can further improve the model performance on the ILSO dataset.

Multi-level feature fusion is a fundamental topic in computer vision. It has been exploited to detect, segment and classify objects at various scales. When multi-level features meet multi-modal cues, the optimal feature aggregation and multi-modal learning strategy become a hot potato. In this paper, we leverage the inherent multi-modal and multi-level nature of RGB-D salient object detection to devise a novel Bifurcated Backbone Strategy Network (BBS-Net). Our architecture, is simple, efficient, and backbone-independent. In particular, first, we propose to regroup the multi-level features into teacher and student features using a bifurcated backbone strategy (BBS). Second, we introduce a depth-enhanced module (DEM) to excavate informative depth cues from the channel and spatial views. Then, RGB and depth modalities are fused in a complementary way. Extensive experiments show that BBS-Net significantly outperforms 18 state-of-the-art (SOTA) models on eight challenging datasets under five evaluation measures, demonstrating the superiority of our approach (~4% improvement in S-measure vs . the top-ranked model: DMRA). In addition, we provide a comprehensive analysis on the generalization ability of different RGB-D datasets and provide a powerful training set for future research. The complete algorithm, benchmark results, and post-processing toolbox are publicly available at https://github.com/zyjwuyan/BBS-Net.

Spectral clustering is a quite simple but effective method for solving graph clustering problem. It projects the original data points into a lower dimensional space with spectral embedding, and then relies on an algorithm to obtain the cluster labels. Since it involves eigendecomposition of the graph Laplacian matrix for embedding, spectral clustering has high time complexity and is not able to process large scale data. The performance of spectral clustering is also limited by a post-processing algorithm such as kmeans. To tackle the two issues, we propose a method called Orthogonal and Nonnegative Graph Reconstruction (ONGR) for large scale clustering. The two constraints serve as a structure constraint with which the graph reconstructed by the indicator matrix is structured. The proposed method has linear time complexity with respect to the data size that it mainly needs to implicitly construct a graph and iteratively perform economical singular value decomposition for a small size matrix. Moreover, the interpretability of the indicator matrix is offered due to the nonnegative constraint, and thus our method can provide the cluster labels with no post-processing. The experiments on benchmark datasets show the effectiveness of the proposed scalable clustering method.

Deep learning has been successfully applied to learn local feature descriptors in recent years. However, most of existing methods are supervised methods relying on a large number of labeled training patches, which are also proposed for learning real valued descriptors. In this paper, we propose a novel unsupervised deep learning method for binary descriptor learning. The binary descriptors are much more compact and efficient than the real valued descriptors and unsupervised leaning is highly required in many applications due to its label-free characteristic as the annotations are sometimes expensive to obtain. The core idea of our method is to explore the locality consistency in the descriptor space as well as to distinguish different patches while maintaining the ability to match a patch with its geometric transformed ones. We also give a theorical analysis about the role of batch normalization in learning effective binary descriptors. Benefited from this analysis, there is no need to append two additional losses on minimizing the quantization error and maximizing the entropy to the final learning objective like previous works did, thus simplifying our network training. Experiments on four benchmarks demonstrate that the proposed method is able to learn binary descriptors significantly outperforming previous unsupervised binary descriptors, even superior to most supervised ones. Especially, it obtains 21.2% of improvement on the UBC Phototour dataset, and 19.8%, 26.7%, 26.0% of improvements for patch verification, matching, retrieval tasks respectively on the HPatches dataset compared to the previous best unsupervised method.

As an interesting and important problem in computer vision, learning-based video saliency detection aims to discover the visually interesting regions in a video sequence. Capturing the information within frame and between frame at different aspects (such as spatial contexts, motion information, temporal consistency across frames, and multiscale representation) is important for this task. A key issue is how to jointly model all these factors within a unified data-driven scheme in an end-to-end fashion. In this article, we propose an end-to-end spatiotemporal deep video saliency detection approach, which captures the information on spatial contexts and motion characteristics. Furthermore, it encodes the temporal consistency information across the consecutive frames by implementing a convolutional long short-term memory (Conv-LSTM) model. In addition, the multiscale saliency properties for each frame are adaptively integrated for final saliency prediction in a collaborative feature-pyramid way. Finally, the proposed deep learning approach unifies all the aforementioned parts into an end-to-end joint deep learning scheme. Experimental results demonstrate the effectiveness of our approach in comparison with the state-of-the-art approaches.

Multi-view classification optimally integrates various features from different views to improve classification tasks. Though most of the existing works demonstrate promising performance in various computer vision applications, we observe that they can be further improved by sufficiently utilizing complementary view-specific information, deep interactive information between different views, and the strategy of fusing various views. In this work, we propose a novel multi-view learning framework that seamlessly embeds various view-specific information and deep interactive information and introduces a novel multi-view fusion strategy to make a joint decision during the optimization for classification. Specifically, we utilize different deep neural networks to learn multiple view-specific representations, and model deep interactive information through a shared interactive network using the cross-correlations between attributes of these representations. After that, we adaptively integrate multiple neural networks by flexibly tuning the power exponent of weight, which not only avoids the trivial solution of weight but also provides a new approach to fuse outputs from different deterministic neural networks. Extensive experiments on several public datasets demonstrate the rationality and effectiveness of our method.

Neuroimaging techniques have shown increasing power to understand the neuropathology of brain disorders. Multimodal brain imaging data carry distinct but complementary information and thus could depict brain disorders comprehensively. To deepen our understanding, it is essential to investigate the intrinsic associations among multiple modalities. To date, the pairwise correlations between imaging data captured by different imaging modalities have been well studied, leaving formidable challenges to identify high-order associations. In this paper, we first propose a new sparse tensor canonical correlation analysis (STCCA) with feature selection to analyze the complex high-order relationships among multimodal brain imaging data. In addition, we find that methods for identifying pairwise associations and high-order associations have complementary advantages, providing a sound reason to fuse them. Therefore, we further propose an improved STCCA (STCCA+) which integrates STCCA and sparse multiple CCA (SMCCA) to fully uncover associations among multiple imaging modalities. The proposed STCCA+detects equivalent association levels among multimodal imaging data compared to SMCCA. Most importantly, both STCCA and STCCA+yield modality-consistent imaging markers and modality-specific ones, assuring a better and meaningful feature selection capability. Finally, the identified imaging markers and their high-order correlations could form a comprehensive indication of brain disorders, showing their promise in high-order multimodal brain imaging analysis.

Saliency detection on RGB-D images is receiving more and more research interests recently. Previous models adopt the early fusion or the result fusion scheme to fuse the input RGB and depth data or their saliency maps, which incur the problem of distribution gap or information loss. Some other models use the feature fusion scheme but are limited by the linear feature fusion methods. In this paper, we propose to fuse attention learned in both modalities. Inspired by the Non-local model, we integrate the self-attention and each other's attention to propagate long-range contextual dependencies, thus incorporating multi-modal information to learn attention and propagate contexts more accurately. Considering the reliability of the other modality's attention, we further propose a selection attention to weight the newly added attention term. We embed the proposed attention module in a two-stream CNN for RGB-D saliency detection. Furthermore, we also propose a residual fusion module to fuse the depth decoder features into the RGB stream. Experimental results on seven benchmark datasets demonstrate the effectiveness of the proposed model components and our final saliency model. Our code and saliency maps are available at https://github.com/nnizhang/S2MA.

Three-dimensional (3D) object classification is widely involved in various computer vision applications, e.g., autonomous driving, simultaneous localization and mapping, which has attracted lots of attention in the committee. However, solving 3D object classification by directly employing the 3D convolutional neural networks (CNNs) generally suffers from high computational cost. Besides, existing view-based methods cannot better explore the content relationships between views. To this end, this work proposes a novel multi-view framework by jointly using multiple 2D-CNNs to capture discriminative information with relationships as well as a new multi-view loss fusion strategy, in an end-to-end manner. Specifically, we utilize multiple 2D views of a 3D object as input and integrate the intra-view and inter-view information of each view through the view-specific 2D-CNN and a series of modules (outer product, view pair pooling, 1D convolution, and fully connected transformation). Furthermore, we design a novel view ensemble mechanism that selects several discriminative and informative views to jointly infer the category of a 3D object. Extensive experiments demonstrate that the proposed method is able to outperform current state-of-the-art methods on 3D object classification. More importantly, this work provides a new way to improve 3D object classification from the perspective of fully utilizing well-established 2D-CNNs.

MOTIVATION: Brain imaging genetics studies the complex associations between genotypic data such as single nucleotide polymorphisms (SNPs) and imaging quantitative traits (QTs). The neurodegenerative disorders usually exhibit the diversity and heterogeneity, originating from which different diagnostic groups might carry distinct imaging QTs, SNPs and their interactions. Sparse canonical correlation analysis (SCCA) is widely used to identify bi-multivariate genotype-phenotype associations. However, most existing SCCA methods are unsupervised, leading to an inability to identify diagnosis-specific genotype-phenotype associations. RESULTS: In this article, we propose a new joint multitask learning method, named MT-SCCALR, which absorbs the merits of both SCCA and logistic regression. MT-SCCALR learns genotype-phenotype associations of multiple tasks jointly, with each task focusing on identifying one diagnosis-specific genotype-phenotype pattern. Meanwhile, MT-SCCALR cannot only select relevant SNPs and imaging QTs for each diagnostic group alone, but also allows the selection of those shared by multiple diagnostic groups. We derive an efficient optimization algorithm whose convergence to a local optimum is guaranteed. Compared with two state-of-the-art methods, MT-SCCALR yields better or similar canonical correlation coefficients and classification performances. In addition, it owns much better discriminative canonical weight patterns of great interest than competitors. This demonstrates the power and capability of MTSCCAR in identifying diagnostically heterogeneous genotype-phenotype patterns, which would be helpful to understand the pathophysiology of brain disorders. AVAILABILITY AND IMPLEMENTATION: The software is publicly available at https://github.com/dulei323/MTSCCALR. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

Object localization and segmentation in weakly labeled videos are two interesting yet challenging tasks. Models built for simultaneous object localization and segmentation have been explored in the conventional fully supervised learning scenario to boost the performance of each task. However, none of the existing works has attempted to jointly learn object localization and segmentation models under weak supervision. To this end, we propose a joint learning framework called Self-Paced Fine-Tuning Network (SPFTN) for localizing and segmenting objects in weakly labelled videos. Learning the deep model jointly for object localization and segmentation under weak supervision is very challenging as the learning process of each single task would face serious ambiguity issue due to the lack of bounding-box or pixel-level supervision. To address this problem, our proposed deep SPFTN model is carefully designed with a novel multi-task self-paced learning objective, which leverages the task-specific prior knowledge and the knowledge that has been already captured to infer the confident training samples for each task. By aggregating the confident knowledge from each single task to mine reliable patterns and learning deep feature representation for both tasks, the proposed learning framework can address the ambiguity issue under weak supervision with simple optimization. Comprehensive experiments on the large-scale YouTube-Objects and DAVIS datasets demonstrate that the proposed approach achieves superior performance when compared with other state-of-the-art methods and the baseline networks/models.

Recently, the research field of salient object detection is undergoing a rapid and remarkable development along with the wide usage of deep neural networks. Being trained with a large number of images annotated with strong pixel-level ground-truth masks, the deep salient object detectors have achieved the state-of-the-art performance. However, it is expensive and time-consuming to provide the pixel-level ground-truth masks for each training image. To address this problem, this paper proposes one of the earliest frameworks to learn deep salient object detectors without requiring any human annotation. The supervisory signals used in our learning framework are generated through a novel supervision synthesis scheme, in which the key insights are "knowledge source transition" and "supervision by fusion". Specifically, in the proposed learning framework, both the external knowledge source and the internal knowledge source are explored dynamically to provide informative cues for synthesizing supervision required in our approach, while a two-stream fusion mechanism is also established to implement the supervision synthesis process. Comprehensive experiments on four benchmark datasets demonstrate that the deep salient object detector trained by our newly proposed learning framework often works well without requiring any human annotated masks, which even approaches to its upper-bound obtained under the fully supervised learning fashion (within only 3 percent performance gap). Besides, we also apply the salient object detector learnt with our annotation-free learning framework to assist the weakly supervised semantic segmentation task, which demonstrates that our approach can also alleviate the heavy supplementary supervision required in the existing weakly supervised semantic segmentation framework.

Although video enhancement and evaluation have been studied for many years, they are still challenging due to the evolutions of video acquiring and processing techniques. While the development of deep learning is undoubtedly exciting and has demonstrated its superior performance in a variety of applications, it is important to further investigate advanced technologies and solutions to bring seemingly endless possibilities for video enhancement and evaluation. This Special Section intends to collect some recent solutions toward video enhancement and evaluation.

Most existing binary descriptors are only based on intensity and simply compare averaged intensities of one sample pair to obtain binary output. To address this issue, we propose a novel ring-sampling pattern based binary descriptor encoding both intensity and gradient. For intensity coding, a ring-sampling pattern is presented to define a number of sample points and their neighboring points given an interest point. The intensity difference between two sample points is obtained by performing the comparisons of the intensities of their corresponding neighboring points directly. Further, a majority based voting strategy is employed to obtain compact representation (1 bit or 3 bits) based on all these intensity difference among neighboring points. As for gradient coding, the gradient orientation histogram is computed for each sample point, and the gradient difference of two sample points is obtained by comparing the gradient magnitude on each orientation bin. The raw binary descriptor is constructed by concentrating the intensity and gradient differences of all the sample pairs, and two feature selection strategies are proposed to obtain the final compact descriptor, named as Features Combined Binary Descriptor based on Voted Ring-Sampling Pattern (BDVRP). The experimental results on the tasks of object recognition and image matching demonstrate the superiority and the effectiveness of the proposed descriptor, with comparison to the state-of-the-art hand-crafted binary descriptors.

Although deep features have achieved the state-of-the-art performance in action recognition recently, the hand-crafted shallow features still play a critical role in characterizing human actions for taking advantage of visual contents in an intuitive way such as edge features. Therefore, the shallow features can serve as auxiliary visual cues supplementary to deep representations. In this paper, we propose a discriminative subspace learning model (DSLM) to explore the complementary properties between the hand-crafted shallow feature representations and the deep features. As for the RGB action recognition, this is the first work attempting to mine multi-level feature complementaries by the multi-view subspace learning scheme. To sufficiently capture the complementary information among heterogeneous features, we construct the DSLM by integrating the multi-view reconstruction error and classification error into an unified objective function. To be specific, we first use Fisher Vector to encode improved dense trajectories (iDT+FV) for shallow representations and two-stream convolutional neural network models (T-CNN) for generating deep features. Moreover, the presented DSLM algorithm projects multi-level features onto a shared discriminative subspace with the complementary information and discriminating capacity simultaneously incorporated. Finally, the action types of test samples are identified by the margins from the learned compact representations to the decision boundary. The experimental results on three datasets demonstrate the effectiveness of the proposed method.

Weakly supervised learning has been attracting much attention due to its broad applications, which only requires image-level annotations to indicate whether there exist objects in the images. Currently, most of the existing weakly supervised object detection (WSOD) methods are inclined to seek only one top-scoring object instance per image from noisy proposals to train the corresponding object detector. However, more than one same-class instances often exist in the large-scale, cluttered remote sensing images. Thus, selecting only one top-scoring proposal usually results in highlighting the most representative part of an object rather than the whole object, which may cause learning a suboptimal object detector by losing much important information. To address this problem, a novel end-to-end progressive contextual instance refinement (PCIR) method is proposed to perform WSOD. Specifically, a dual-contextual instance refinement (DCIR) strategy is designed to divert the focus of the detection network from the local distinct part to the whole object and further to other potential instances by leveraging both local and global context information. Benefiting from DCIR, a progressive proposal self-pruning (PPSP) strategy is further developed to mitigate the influence of the complex background by dynamically rejecting the negative training proposals. Comprehensive experiments on the challenging NWPU VHR-10.v2 and DIOR data sets clearly demonstrate that the proposed method can significantly boost the detection accuracy compared with the state of the arts.

Despite significant efforts made so far for Weakly Supervised Object Detection (WSOD), proposal generation and proposal selection are still two major challenges. In this paper, we focus on addressing the two challenges by generating and selecting high-quality proposals. To be specific, for proposal generation, we combine selective search and a Gradient-weighted Class Activation Mapping (Grad-CAM) based technique to generate more proposals having higher Intersection-Over-Union (IOU) with ground truth boxes than those obtained by greedy search approaches, which can better envelop the entire objects. As regards proposal selection, for each object class, we choose as many confident positive proposals as possible and meanwhile only select class-specific hard negatives to focus training on more discriminative negative proposals by up-weighting their losses, which can make training more effective. The proposed proposal generation and proposal selection approaches are generic and thus can be broadly applied to many WSOD methods. In this work, we unify them into the framework of Online Instance Classifier Refinement (OICR). Experimental results on the PASCAL VOC 2007 and 2012 datasets and MS COCO dataset demonstrate that our method significantly improves the baseline method OICR by large margins (13.4% mAP and 11.6% CorLoc gains on the VOC 2007 dataset, 15.0% mAP and 8.9% CorLoc gains on the VOC 2012 dataset, and 6.4% mAP and 5.0% CorLoc gains on the COCO dataset) and achieves the state-of-the-art results compared with existing methods.

Existing saliency models typically incorporate contexts holistically. However, for each pixel, usually only part of its context region contributes to saliency prediction, while other parts are likely either noise or distractions. In this paper, we propose a novel pixel-wise contextual attention network (PiCANet) to selectively attend to informative context locations at each pixel. The proposed PiCANet generates an attention map over the contextual region of each pixel and construct attentive contextual features via selectively incorporating the features of useful context locations. We present three formulations of the PiCANet via embedding the pixel-wise contextual attention mechanism into the pooling and convolution operations with attending to global or local contexts. All the three models are fully differentiable and can be integrated with convolutional neural networks with joint training. In this work, we introduce the proposed PiCANets into a U-Net model for salient object detection. The generated global and local attention maps can learn to incorporate global contrast and regional smoothness, which help localize and highlight salient objects more accurately and uniformly. Experimental results show that the proposed PiCANets perform effectively for saliency detection against the state-of-the-art methods. Furthermore, we demonstrate the effectiveness and generalization ability of the PiCANets on semantic segmentation and object detection with improved performance.

Most of the current action localization methods follow an anchor-based pipeline: depicting action instances by pre-defined anchors, learning to select the anchors closest to the ground truth, and predicting the confidence of anchors with refinements. Pre-defined anchors set prior about the location and duration for action instances, which facilitates the localization for common action instances but limits the flexibility for tackling action instances with drastic varieties, especially for extremely short or extremely long ones. To address this problem, this paper proposes a novel anchor-free action localization module that assists action localization by temporal points. Specifically, this module represents an action instance as a point with its distances to the starting boundary and ending boundary, alleviating the pre-defined anchor restrictions in terms of action localization and duration. The proposed anchor-free module is capable of predicting the action instances whose duration is either extremely short or extremely long. By combining the proposed anchor-free module with a conventional anchor-based module, we propose a novel action localization framework, called A2Net. The cooperation between anchor-free and anchor-based modules achieves superior performance to the state-of-the-art on THUMOS14 (45.5% vs. 42.8%). Furthermore, comprehensive experiments demonstrate the complementarity between the anchor-free and the anchor-based module, making A2Net simple but effective.

Brain tumor segmentation, which aims at segmenting the whole tumor area, enhancing tumor core area, and tumor core area from each input multi-modality bioimaging data, has received considerable attention from both academia and industry. However, the existing approaches usually treat this problem as a common semantic segmentation task without taking into account the underlying rules in clinical practice. In reality, physicians tend to discover different tumor areas by weighing different modality volume data. Also, they initially segment the most distinct tumor area, and then gradually search around to find the other two. We refer to the first property as the task-modality structure while the second property as the task-task structure, based on which we propose a novel task-structured brain tumor segmentation network (TSBTS net). Specifically, to explore the task-modality structure, we design a modality-aware feature embedding mechanism to infer the important weights of the modality data during network learning. To explore the tasktask structure, we formulate the prediction of the different tumor areas as conditional dependency sub-tasks and encode such dependency in the network stream. Experiments on BraTS benchmarks show that the proposed method achieves superior performance in segmenting the desired brain tumor areas while requiring relatively lower computational costs, compared to other state-of-the-art methods and baseline models.

With the explosive growth of data, the multi-view data is widely used in many fields, such as data mining, machine learning, computer vision, and so on. Because such data always has a complex structure, i.e., many categories, many perspectives of description and high dimension, how to formulate an accurate and reliable framework for the multi-view classification is a very challenging task. In this paper, we propose a novel multi-view classification method by using multiple multi-class Support Vector Machines (SVMs) with a novel collaborative strategy. Here, each multi-class SVM embeds the scaling factor to renewedly adjust the weight allocation of all features, which is beneficial to highlight more important and discriminative features. Furthermore, we adopt the decision function values to integrate multiple multi-class learners and introduce the confidence score across multiple classes to determine the final classification result. In addition, through a series of the mathematical deduction, we bridge the proposed model with the solvable problem and solve it through an alternating iteration optimization method. We evaluate the proposed method on several image and face datasets, and the experimental results demonstrate that our proposed method performs better than other state-of-the-art learning algorithms.

Brain imaging genetics becomes more and more important in brain science, which integrates genetic variations and brain structures or functions to study the genetic basis of brain disorders. The multi-modal imaging data collected by different technologies, measuring the same brain distinctly, might carry complementary information. Unfortunately, we do not know the extent to which the phenotypic variance is shared among multiple imaging modalities, which further might trace back to the complex genetic mechanism. In this paper, we propose a novel dirty multi-task sparse canonical correlation analysis (SCCA) to study imaging genetic problems with multi-modal brain imaging quantitative traits (QTs) involved. The proposed method takes advantages of the multi-task learning and parameter decomposition. It can not only identify the shared imaging QTs and genetic loci across multiple modalities, but also identify the modality-specific imaging QTs and genetic loci, exhibiting a flexible capability of identifying complex multi-SNP-multi-QT associations. Using the state-of-the-art multi-view SCCA and multi-task SCCA, the proposed method shows better or comparable canonical correlation coefficients and canonical weights on both synthetic and real neuroimaging genetic data. In addition, the identified modality-consistent biomarkers, as well as the modality-specific biomarkers, provide meaningful and interesting information, demonstrating the dirty multi-task SCCA could be a powerful alternative method in multi-modal brain imaging genetics.

Weakly supervised object detection (WSOD) is an interesting yet challenging task in the computer vision community. The core is to discover the image regions that contain the complete object instances under the image-level supervision. Existing works usually solve this problem via a proposal selection strategy, which selects the most discriminative box regions from the weakly labeled training images. However, these regions usually only contain the discriminative object parts rather than the complete object instances. To address this problem, this article proposes to learn a searching-agent to gradually mine desirable object regions under a region searching paradigm, where we formulate the searching process as a Markov decision process and learn the searching-agent under a deep reinforcement learning framework. To learn such a searching-agent under the weak supervision, we extract the pseudo-complete object regions and the corresponding local discriminative object parts and introduce the obtained pseudo-target-part training pairs into the reinforcement learning process of the search-agent. This learning strategy has twofold advantages: 1) it can mimic the searching process to reveal complete object regions from a certain discriminative part of the object under the weak supervision and 2) it will not suffer from the learning difficulty arise from the long-action sequence that happens when searching from the entire image range. Comprehensive experiments on benchmark data sets demonstrate that by integrating the learned searching-agent with the existing WSOD method, we can achieve better performance than the other state-of-the-art and baseline methods.

With the advances of convolutional neural networks (CNNs), the accuracy of remote sensing image scene classification has been greatly boosted thanks to the powerful features extracted through CNNs. Although significant success has been achieved, most of existing methods are dominated by the use of fully-connected CNN features. This paper focuses on the performance comparison of two kinds of novel pooling strategies, including generalized max pooling (GMP) and taskdriven pooling (TDP), for remote sensing image scene classification. To this end, an off-the-shelf CNN model is used as backbone network to extract multi-scale convolutional features. Then, GMP and TDP are respectively adopted to obtain globally pooled features. Finally, scene classification is performed with support vector machine (SVM). In the experiment, we evaluate the performance of these two kinds of pooling schemes on a widely-used scene classification benchmark data set. The experimental results show that (i) using pooled CNN convolutional features can obtain better results than using fully-connected CNN features and (ii) TDP is slightly better than GMP.

Recently, deep learning especially convolutional neural networks (CNNs) has huge great success for remote sensing image scene classification. However, global CNN features still lack geometric invariance for addressing the problem of large intra-class variations and so are not optimal for scene classification. In this paper, we introduce a new feature representation for scene classification, named region response ranking (3R) feature representations by using off-the-shelf CNN models. Specifically, by considering each cube pixel of a certain convolutional feature map as one image region, we jointly train a class-specific support vector machine (SVM) base classifier and a decision function for each scene class. The base classifier is used to generate 3R feature by reordering the SVM responses of all image regions in descending order and the decision function is used for classification with 3R feature representations. Comprehensive evaluations on the publicly available NWPU-RESISC45 data set and comparisons with state-of-the-art methods demonstrate that the proposed 3R feature is effective for remote sensing image scene classification.1

Object detection in very high resolution (VHR) optical remote sensing images is a fundamental yet challenging problem for the field of remote sensing image analysis. The detection performance is heavily dependent on the representation capability of the extracted features. Recently, convolutional neural networks (CNNs) have made a breakthrough for various applications in nature images. However, it is problematic to directly apply CNN to perform object detection in VHR optical remote sensing images due to the problem of object rotation variations. To address this issue, a novel rotation invariant probabilistic Latent Semantic Analysis (RI-pLSA) model is proposed to learn latent semantic representations for object detection. This is achieved by imposing a rotation-invariant regularization term on the objective function of pLSA to enforce the learned representation from all rotations of the same sample to be as consistent as possible. Additionally, the proposed RI-pLSA model takes the CNN features as input, which generates more powerful semantic representation for object detection. Comprehensive experiments on a publicly available ten-class object detection dataset demonstrate the superiority and effectiveness of our method compared with state-of-the-arts.

Scene classification of high resolution remote sensing (HRRS) images is a fundamental yet challenging problem for remote sensing image analysis. In this paper, we focus on tackling the problem of HRSS scene classification using a small pool of unlabeled images and only a few labeled images per category, namely, few-shot scene classification (FSSC), which is more challenging than common scene classification task. The key challenge arises from selecting trustworthy samples from the pool of unlabeled images that have high confidence. To address this challenge, a novel local manifold constrained self-paced deep learning method is proposed. Specifically, the model is learned by gradually selecting easy samples from the pool of unlabeled images, assigning them with pseudo-labels and further adopting them with labeled images as the new training set. In addition, a local manifold constraint is introduced to enforce that the pseudo-labels assigned by the initial model should be consistent with the local manifold of the labeled samples. In such way, the confidence of the selecting samples is increased and is beneficial to train more robust classifier. Experimental results on a publicly available large scale NWPU-RESISC45 data set demonstrated the effectiveness of our method in achieving competitive performance while significantly reducing manually labeled cost.

The local binary pattern (LBP) and its variants have shown the effectiveness in texture images classification, face recognition, and other applications. However, most of these LBP methods only focus on the histogram of LBP patterns and ignore the spatial contextual information between LBP patterns. In this paper, we propose a 2D-LBP method which uses a sliding window to count the weighted occurrence number of the rotation invariant uniform LBP pattern pairs to obtain the spatial contextual information. The multi-resolution 2D-LBP features can also be obtained when the radius of 2D-LBP is changed. At last, a two-stage classifier which acts as an ensemble learning step is followed to achieve an accurate classification by combining the predictions on each 2D-LBP with single resolution. Theoretical proof shows that the proposed 2D-LBP is a general framework and can be integrated on other LBP variants to derive new feature extraction methods. Experimental results show that, the proposed method achieves 99.71%, 97.09%, 98.48%, and 49.00% classification accuracy on the public “Brodatz,” “CUReT,” “UIUC,” and “FMD” texture image databases, respectively. Compared with the original LBP and its variants, the proposed method obtains higher classification accuracy under different cases, and simultaneously owns shorter time complexity.

For alleviating the human labor associated with annotating the training data for learning object detectors, recent research has focused on semi-supervised object detection (SSOD) and weakly supervised object detection (WSOD) approaches. In SSOD, instead of annotating all the instances in the whole training set, people only need to annotate the part of the training instances using bounding boxes. In WSOD, people need to annotate the image-level tags on all training images to indicate the object categories contained by the corresponding images since more detailed bounding box annotations are no longer needed. Along this line of research, this paper makes a further step to alleviate the human labor in annotating training data, leading to the problem of object detection with semi-annotated weak labels (ODSAWLs). Instead of labeling image-level tags on all training images, ODSAWL only needs the image-level tags for a small portion of the training images, and then, the object detectors can be learned from a small portion of the weakly-labeled training images and from the remaining unlabeled training images. To address such a challenging problem, this paper proposes a cross model co-training framework that collaborates an object localizer and a tag generator in an alternative optimization procedure. Specifically, during the learning procedure, these two (deep) models can transfer the needed knowledge (including labels and visual patterns) between each other. The whole learning procedure is accomplished in a few stages under the guidance of a progressive learning curriculum. To demonstrate the effectiveness of the proposed approach, we implement the comprehensive experiments on three benchmark datasets, where the obtained experimental results are quite encouraging. Notably, by using only about 15% weakly labeled training images, the proposed approach can effectively approach, or even outperform, the state-of-the-art WSOD methods.

Effective 3-D local features are significant elements for 3-D shape analysis. Existing hand-crafted 3-D local descriptors are effective but usually involve intensive human intervention and prior knowledge, which burdens the subsequent processing procedures. An alternative resorts to the unsupervised learning of features from raw 3-D representations via popular deep learning models. However, this alternative suffers from several significant unresolved issues, such as irregular vertex topology, arbitrary mesh resolution, orientation ambiguity on the 3-D surface, and rigid and slightly nonrigid transformation invariance. To tackle these issues, we propose an unsupervised 3-D local feature learning framework based on a novel permutation voxelization strategy to learn high-level and hierarchical 3-D local features from raw 3-D voxels. Specifically, the proposed strategy first applies a novel voxelization which discretizes each 3-D local region with irregular vertex topology and arbitrary mesh resolution into regular voxels, and then, a novel permutation is applied to permute the voxels to simultaneously eliminate the effect of rotation transformation and orientation ambiguity on the surface. Based on the proposed strategy, the permuted voxels can fully encode the geometry and structure of each local region in regular, sparse, and binary vectors. These voxel vectors are highly suitable for the learning of hierarchical common surface patterns by stacked sparse autoencoder with hierarchical abstraction and sparse constraint. Experiments are conducted on three aspects for evaluating the learned local features: 1) global shape retrieval; 2) partial shape retrieval; and 3) shape correspondence. The experimental results show that the learned local features outperform the other state-of-the-art 3-D shape descriptors.

As one of the fundamental research topics in remote sensing image analysis, hyperspectral image (HSI) classification has been extensively studied so far. However, how to discriminatively learn a low-dimensional feature space, in which the mapped features have small within-class scatter and big between-class separation, is still a challenging problem. To address this issue, this paper proposes an effective framework, named compact and discriminative stacked autoencoder (CDSAE), for HSI classification. The proposed CDSAE framework comprises two stages with different optimization objectives, which can learn discriminative low-dimensional feature mappings and train an effective classifier progressively. First, we impose a local Fisher discriminant regularization on each hidden layer of stacked autoencoder (SAE) to train discriminative SAE (DSAE) by minimizing reconstruction error. This stage can learn feature mappings, in which the pixels from the same land-cover class are mapped as nearly as possible and the pixels from different land-cover categories are separated by a large margin. Second, we learn an effective classifier and meanwhile update DSAE with a local Fisher discriminant regularization being embedded on the top of feature representations. Moreover, to learn a compact DSAE with as small number of hidden neurons as possible, we impose a diversity regularization on the hidden neurons of DSAE to balance the feature dimensionality and the feature representation capability. The experimental results on three widely-used HSI data sets and comprehensive comparisons with existing methods demonstrate that our proposed method is effective.

The performance of object detection has recently been significantly improved due to the powerful features learnt through convolutional neural networks (CNNs). Despite the remarkable success, there are still several major challenges in object detection, including object rotation, within-class diversity, and between-class similarity, which generally degenerate object detection performance. To address these issues, we build up the existing state-of-the-art object detection systems and propose a simple but effective method to train rotation-invariant and Fisher discriminative CNN models to further boost object detection performance. This is achieved by optimizing a new objective function that explicitly imposes a rotation-invariant regularizer and a Fisher discrimination regularizer on the CNN features. Specifically, the first regularizer enforces the CNN feature representations of the training samples before and after rotation to be mapped closely to each other in order to achieve rotation-invariance. The second regularizer constrains the CNN features to have small within-class scatter but large between-class separation. We implement our proposed method under four popular object detection frameworks, including region-CNN (R-CNN), Fast R- CNN, Faster R- CNN, and R- FCN. In the experiments, we comprehensively evaluate the proposed method on the PASCAL VOC 2007 and 2012 data sets and a publicly available aerial image data set. Our proposed methods outperform the existing baseline methods and achieve the state-of-the-art results.

Learning 3D global features by aggregating multiple views is important. Pooling is widely used to aggregate views in deep learning models. However, pooling disregards a lot of content information within views and the spatial relationship among the views, which limits the discriminability of learned features. To resolve this issue, 3D to Sequential Views (3D2SeqViews) is proposed to more effectively aggregate the sequential views using convolutional neural networks with a novel hierarchical attention aggregation. Specifically, the content information within each view is first encoded. Then, the encoded view content information and the sequential spatiality among the views are simultaneously aggregated by the hierarchical attention aggregation, where view-level attention and class-level attention are proposed to hierarchically weight sequential views and shape classes. View-level attention is learned to indicate how much attention is paid to each view by each shape class, which subsequently weights sequential views through a novel recursive view integration. Recursive view integration learns the semantic meaning of view sequence, which is robust to the first view position. Furthermore, class-level attention is introduced to describe how much attention is paid to each shape class, which innovatively employs the discriminative ability of the fine-tuned network. 3D2SeqViews learns more discriminative features than the state-of-the-art, which leads to the outperforming results in shape classification and retrieval under three large-scale benchmarks.

Learning 3D global features by aggregating multiple views has been introduced as a successful strategy for 3D shape analysis. In recent deep learning models with end-to-end training, pooling is a widely adopted procedure for view aggregation. However, pooling merely retains the max or mean value over all views, which disregards the content information of almost all views and also the spatial information among the views. To resolve these issues, we propose Sequential Views To Sequential Labels (SeqViews2SeqLabels) as a novel deep learning model with an encoder-decoder structure based on recurrent neural networks (RNNs) with attention. SeqViews2SeqLabels consists of two connected parts, an encoder-RNN followed by a decoder-RNN, that aim to learn the global features by aggregating sequential views and then performing shape classification from the learned global features, respectively. Specifically, the encoder-RNN learns the global features by simultaneously encoding the spatial and content information of sequential views, which captures the semantics of the view sequence. With the proposed prediction of sequential labels, the decoder-RNN performs more accurate classification using the learned global features by predicting sequential labels step by step. Learning to predict sequential labels provides more and finer discriminative information among shape classes to learn, which alleviates the overfitting problem inherent in training using a limited number of 3D shapes. Moreover, we introduce an attention mechanism to further improve the discriminative ability of SeqViews2SeqLabels. This mechanism increases the weight of views that are distinctive to each shape class, and it dramatically reduces the effect of selecting the first view position. Shape classification and retrieval results under three large-scale benchmarks verify that SeqViews2SeqLabels learns more discriminative global features by more effectively aggregating sequential views than state-of-the-art methods.

Video summarization is a challenging task, mainly due to the difficulties in learning complicated semantic structural relations between videos and summaries. In this paper, we present a novel supervised video summarization scheme based on threestage deep neural networks. The scheme takes a divide-andconquer strategy to resolve the complicated task of 3D video summarization into a set of easy and flexible computational subtasks, and then to sequentially perform 2D CNNs, 1D CNNs, and LSTM to address the subtasks in an hierarchical fashion. The hierarchical modeling of spatio-temporal structure leads to high performance and efficiency. In addition, we propose a simple but effective user-ranking method to cope with the labeling subjectivity problem of user-created video summarization, leading to the labeling quality refinement for robust supervised learning. Experimental results show that our approach outperforms the state-of-the-art video summarization methods on two benchmark datasets.

Visual attention is a dynamic process of scene exploration and information acquisition. However, existing research on attention modeling has concentrated on estimating static salient locations. In contrast, dynamic attributes presented by saccade have not been well explored in previous attention models. In this paper, we address the problem of saccadic scanpath prediction by introducing an iterative representation learning framework. Within the framework, saccade can be interpreted as an iterative process of predicting one fixation according to the current representation and updating the representation based on the gaze shift. In the predicting phase, we propose a Bayesian definition of saccade to combine the influence of perceptual residual and spatial location on the selection of fixations. In implementation, we compute the representation error of an autoencoder-based network to measure perceptual residuals of each area. Simultaneously, we integrate saccade amplitude and center-weighted mechanism to model the influence of spatial location. Based on estimating the influence of two parts, the final fixation is defined as the point with the largest posterior probability of gaze shift. In the updating phase, we update the representation pattern for the subsequent calculation by retraining the network with samples extracted around the current fixation. In the experiments, the proposed model can replicate the fundamental properties of psychophysics in visual search. In addition, it can achieve superior performance on several benchmark eye-tracking data sets.

For decades, task functional magnetic resonance imaging has been a powerful noninvasive tool to explore the organizational architecture of human brain function. Researchers have developed a variety of brain network analysis methods for task fMRI data, including the general linear model, independent component analysis, and sparse representation methods. However, these shallow models are limited in faithful reconstruction and modeling of the hierarchical and temporal structures of brain networks, as demonstrated in more and more studies. Recently, recurrent neural networks (RNNs) exhibit great ability of modeling hierarchical and temporal dependence features in the machine learning field, which might be suitable for task fMRI data modeling. To explore such possible advantages of RNNs for task fMRI data, we propose a novel framework of a deep recurrent neural network (DRNN) to model the functional brain networks from task fMRI data. Experimental results on the motor task fMRI data of Human Connectome Project 900 subjects release demonstrated that the proposed DRNN can not only faithfully reconstruct functional brain networks, but also identify more meaningful brain networks with multiple time scales which are overlooked by traditional shallow models. In general, this work provides an effective and powerful approach to identifying functional brain networks at multiple time scales from task fMRI data.

Sparse regression based feature selection method has been extensively investigated these years. However, because it has a non-convex constraint, i.e., $\ell _{2,0}$ℓ2,0-norm constraint, this problem is very hard to solve. In this paper, unlike most of the other methods which only solve its slack version by introducing sparsity regularization into objective function forcibly, a novel framework is proposed by us to solve the original $\ell _{2,0}$ℓ2,0-norm constrained sparse regression based feature selection problem. We transform our objective function into Linear Discriminant Analysis (LDA) by using a new label coding method, thus enabling our model to calculate the ratio of inter-class scatter to intra-class scatter of features which is the most widely used feature discrimination evaluation metric. According to that ratio, features can be selected by a simple sorting method. The projection gradient descent algorithm is introduced to further improve the performance of our algorithm by using the solution obtained before as its initial solution. This ensures the stability of this iterative algorithm. We prove that the proposed method can get the global optimal solution of this non-convex problem when all features are statistically independent. For the general case where features are statistically dependent, extensive experiments on six small sample size datasets and one large-scale dataset show that our algorithm has comparable or better classification capability comparing with other eight state-of-the-art feature selection methods by the SVM classifier. We also show that our algorithm can obtain a low loss value, which means the solution of our algorithm can get very close to this NP-hard problem’s real solution. What is more, because we solve the original $\ell _{2,0}$ℓ2,0-norm constrained problem, we avoid the heavy work of tuning the regularization parameter because its meaning is explicit in our method, i.e., the number of selected features. At last, we evaluate the stability of our algorithm from two perspectives, i.e., the objective function values and the selected features, by experiments. From both perspectives, our algorithm shows satisfactory stability performance.

Brain activity is a dynamic combination of different sensory responses and thus brain activity/state is continuously changing over time. However, the brain's dynamical functional states recognition at fast time-scales in task fMRI data have been rarely explored. In this paper, we propose a novel 5-layer deep sparse recurrent neural network (DSRNN) model to accurately recognize the brain states across the whole scan session. Specifically, the DSRNN model includes an input layer, one fully-connected layer, two recurrent layers, and a softmax output layer. The proposed framework has been tested on seven task fMRI data sets of Human Connectome Project. Extensive experiment results demonstrate that the proposed DSRNN model can accurately identify the brain's state in different task fMRI data sets and significantly outperforms other auto-correlation methods or non-temporal approaches in the dynamic brain state recognition accuracy. In general, the proposed DSRNN offers a new methodology for basic neuroscience and clinical research.

In recent years, content-based image retrieval has been of concern because of practical needs on Internet services, especially methods that can improve retrieving speed and accuracy. The SIFT feature is a well-designed local feature. It has mature applications in feature matching and retrieval, whereas the raw SIFT feature is high dimensional, with high storage cost as well as computational cost in feature similarity measurements. Thus, we propose a hashing scheme for fast SIFT feature-based image matching and retrieval. First, a training process of the hashing function involves geometric and topological information being introduced; second, a geometry-enhanced similarity evaluation that considers both the global and details of images in evaluation is explained. Compared with state-of-the-art methods, our method achieves better performance.

For many specific applications in data mining and machine learning, we face explicit or latent size constraint for each cluster that leads to the "balanced clustering" problem. Many existing clustering algorithms perform well in partitioning but fail in producing balanced clusters and preserving the naturally balanced structure of some data. In this paper, we propose a novel balanced clustering framework that flexibly utilizes local and global information of data. First, we propose the global balanced clustering (GBC), in which a global discriminative partitioning model is combined with the minimization of the distribution entropy of data. Then, we show that the proposed GBC can be further used to globally regularize some widely used local clustering models, so as to transform them into balanced clustering that simultaneously capture local and global data. We apply our global balanced regularization to spectral clustering (SC) and local learning (LL)-based clustering, respectively, and propose another two novel balanced clustering models: the local and global balanced SC (LGB-SC) and LGB-LL. Finding the optimal balanced partition is nondeterministic polynomial-time (NP)-hard in general. We adopt the method of augmented Lagrange multipliers to help optimize our model. Comprehensive experiments on several real world benchmarks demonstrate the advantage of our framework to yield balanced clusters while preserving good clustering quality. Our proposed LGB-SC and LGB-LL also outperform SC and LL as well as other classical clustering methods.

Network representation has been recently exploited for many applications, such as citation recommendation, multi-label classification and link prediction. It learns low-dimensional vector representation for each vertex in networks. Existing network representation methods only focus on incomplete aspects of vertex information (i.e., vertex content, network structure or partial integration), moreover they are commonly designed for homogeneous information networks where all the vertices of a network are of the same type. In this paper, we propose a deep network representation model that integrates network structure and the vertex content information into a unified framework by exploiting generative adversarial network, and represents different types of vertices in the heterogeneous network in a continuous and common vector space. Based on the proposed model, we can obtain heterogeneous bibliographic network representation for efficient citation recommendation. The proposed model also makes personalized citation recommendation possible, which is a new issue that a few papers addressed in the past. When evaluated on the AAN and DBLP datasets, the performance of the proposed heterogeneous bibliographic network based citation recommendation approach is comparable with that of the other network representation based citation recommendation approaches. The results also demonstrate that the personalized citation recommendation approach is more effective than the non-personalized citation recommendation approach.

Video object segmentation is a fundamental yet challenging task in computer vision community. In this paper, we formulate this problem as a Markov Decision Process, where agents are learned to segment object regions under a deep reinforcement learning framework. Essentially, learning agents for segmentation is nontrivial as segmentation is a nearly continuous decision-making process, where the number of the involved agents (pixels or superpixels) and action steps from the seed (super)pixels to the whole object mask might be incredibly huge. To overcome this difficulty, this paper simplifies the learning of segmentation agents to the learning of a cutting-agent, which only has a limited number of action units and can converge in just a few action steps. The basic assumption is that object segmentation mainly relies on the interaction between object regions and their context. Thus, with an optimal object (box) region and context (box) region, we can obtain the desirable segmentation mask through further inference. Based on this assumption, we establish a novel reinforcement cutting-agent learning framework, where the cutting-agent consists of a cutting-policy network and a cutting-execution network. The former learns policies for deciding optimal object-context box pair, while the latter executes the cutting function based on the inferred object-context box pair. With the collaborative interaction between the two networks, our method can achieve the outperforming VOS performance on two public benchmarks, which demonstrates the rationality of our assumption as well as the effectiveness of the proposed learning framework.

Contexts play an important role in the saliency detection task. However, given a context region, not all contextual information is helpful for the final task. In this paper, we propose a novel pixel-wise contextual attention network, i.e., the PiCANet, to learn to selectively attend to informative context locations for each pixel. Specifically, for each pixel, it can generate an attention map in which each attention weight corresponds to the contextual relevance at each context location. An attended contextual feature can then be constructed by selectively aggregating the contextual information. We formulate the proposed PiCANet in both global and local forms to attend to global and local contexts, respectively. Both models are fully differentiable and can be embedded into CNNs for joint training. We also incorporate the proposed models with the U-Net architecture to detect salient objects. Extensive experiments show that the proposed PiCANets can consistently improve saliency detection performance. The global and local PiCANets facilitate learning global contrast and homogeneousness, respectively. As a result, our saliency model can detect salient objects more accurately and uniformly, thus performing favorably against the state-of-the-art methods.

Motion of the human body is the critical cue for understanding and characterizing human behavior in videos. Most existing approaches explore the motion cue using optical flows. However, optical flow usually contains motion on both the interested human bodies and the undesired background. This "noisy" motion representation makes it very challenging for pose estimation and action recognition in real scenarios. To address this issue, this paper presents a novel deep motion representation, called PoseFlow, which reveals human motion in videos while suppressing background and motion blur, and being robust to occlusion. For learning PoseFlow with mild computational cost, we propose a functionally structured spatial-temporal deep network, PoseFlow Net (PFN), to jointly solve the skeleton localization and matching problems of PoseFlow. Comprehensive experiments show that PFN outperforms the state-of-the-art deep flow estimation models in generating PoseFlow. Moreover, PoseFlow demonstrates its potential on improving two challenging tasks in human video analysis: pose estimation and action recognition.

Convolutional neural networks (CNNs) have shown their promise for image classification task. However, global CNN features still lack geometric invariance for addressing the problem of intra-class variations and so are not optimal for multi-label image classification. This paper proposes a new and effective framework built upon CNNs to learn Multi-scale and Discriminative Part Detectors (MsDPD)-based feature representations for multi-label image classification. Specifically, at each scale level, we (i) first present an entropy-rank based scheme to generate and select a set of discriminative part detectors (DPD), and then (ii) obtain a number of DPD-based convolutional feature maps with each feature map representing the occurrence probability of a particular part detector and learn DPD-based features by using a task-driven pooling scheme. The two steps are formulated into a unified framework by developing a new objective function, which jointly trains part detectors incrementally and integrates the learning of feature representations into the classification task. Finally, the multi-scale features are fused to produce the predictions. Experimental results on PASCAL VOC 2007 and VOC 2012 datasets demonstrate that the proposed method achieves better accuracy when compared with the existing state-of-the-art multi-label classification methods.

MOTIVATION: Brain imaging genetics, which studies the linkage between genetic variations and structural or functional measures of the human brain, has become increasingly important in recent years. Discovering the bi-multivariate relationship between genetic markers such as single-nucleotide polymorphisms (SNPs) and neuroimaging quantitative traits (QTs) is one major task in imaging genetics. Sparse Canonical Correlation Analysis (SCCA) has been a popular technique in this area for its powerful capability in identifying bi-multivariate relationships coupled with feature selection. The existing SCCA methods impose either the ℓ1-norm or its variants to induce sparsity. The ℓ0-norm penalty is a perfect sparsity-inducing tool which, however, is an NP-hard problem. RESULTS: In this paper, we propose the truncated ℓ1-norm penalized SCCA to improve the performance and effectiveness of the ℓ1-norm based SCCA methods. Besides, we propose an efficient optimization algorithms to solve this novel SCCA problem. The proposed method is an adaptive shrinkage method via tuning τ. It can avoid the time intensive parameter tuning if given a reasonable small τ. Furthermore, we extend it to the truncated group-lasso (TGL), and propose TGL-SCCA model to improve the group-lasso-based SCCA methods. The experimental results, compared with four benchmark methods, show that our SCCA methods identify better or similar correlation coefficients, and better canonical loading profiles than the competing methods. This demonstrates the effectiveness and efficiency of our methods in discovering interesting imaging genetic associations. AVAILABILITY AND IMPLEMENTATION: The Matlab code and sample data are freely available at http://www.iu.edu/∼shenlab/tools/tlpscca/. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

Co-saliency detection, which focuses on extracting commonly salient objects in a group of relevant images, has been attracting research interest because of its broad applications. In practice, the relevant images in a group may have a wide range of variations, and the salient objects may also have large appearance changes. Such wide variations usually bring about large intra-co-salient objects (intra-COs) diversity and high similarity between COs and background, which makes the co-saliency detection task more difficult. To address these problems, we make the earliest effort to introduce metric learning to co-saliency detection. Specifically, we propose a unified metric learning-based framework to jointly learn discriminative feature representation and co-salient object detector. This is achieved by optimizing a new objective function that explicitly embeds a metric learning regularization term into support vector machine (SVM) training. Here, the metric learning regularization term is used to learn a powerful feature representation that has small intra-COs scatter, but big separation between background and COs and the SVM classifier is used for subsequent co-saliency detection. In the experiments, we comprehensively evaluate the proposed method on two commonly used benchmark data sets. The state-of-the-art results are achieved in comparison with the existing co-saliency detection methods.

Salient object detection from RGB-D images aims to utilize both the depth view and RGB view to automatically localize objects of human interest in the scene. Although a few earlier efforts have been devoted to the study of this paper in recent years, two major challenges still remain: 1) how to leverage the depth view effectively to model the depth-induced saliency and 2) how to implement an optimal combination of the RGB view and depth view, which can make full use of complementary information among them. To address these two challenges, this paper proposes a novel framework based on convolutional neural networks (CNNs), which transfers the structure of the RGB-based deep neural network to be applicable for depth view and fuses the deep representations of both views automatically to obtain the final saliency map. In the proposed framework, the first challenge is modeled as a cross-view transfer problem and addressed by using the task-relevant initialization and adding deep supervision in hidden layer. The second challenge is addressed by a multiview CNN fusion model through a combination layer connecting the representation layers of RGB view and depth view. Comprehensive experiments on four benchmark datasets demonstrate the significant and consistent improvements of the proposed approach over other state-of-the-art methods.

Classifier fusion is an efficient strategy to improve the classification performance for the complex pattern recognition problem. In practice, the multiple classifiers to combine can have different reliabilities and the proper reliability evaluation plays an important role in the fusion process for getting the best classification performance. We propose a new method for classifier fusion with contextual reliability evaluation (CF-CRE) based on inner reliability and relative reliability concepts. The inner reliability, represented by a matrix, characterizes the probability of the object belonging to one class when it is classified to another class. The elements of this matrix are estimated from the -nearest neighbors of the object. A cautious discounting rule is developed under belief functions framework to revise the classification result according to the inner reliability. The relative reliability is evaluated based on a new incompatibility measure which allows to reduce the level of conflict between the classifiers by applying the classical evidence discounting rule to each classifier before their combination. The inner reliability and relative reliability capture different aspects of the classification reliability. The discounted classification results are combined with Dempster-Shafer's rule for the final class decision making support. The performance of CF-CRE have been evaluated and compared with those of main classical fusion methods using real data sets. The experimental results show that CF-CRE can produce substantially higher accuracy than other fusion methods in general. Moreover, CF-CRE is robust to the changes of the number of nearest neighbors chosen for estimating the reliability matrix, which is appealing for the applications.

Wireless capsule endoscopy (WCE) enables clinicians to examine the digestive tract without any surgical operations, at the cost of a large amount of images to be analyzed. The main challenge for automatic computer-aided diagnosis arises from the difficulty of robust characterization of these images. To tackle this problem, a novel discriminative joint-feature topic model (DJTM) with dual constraints is proposed to classify multiple abnormalities in WCE images. We first propose a joint-feature probabilistic latent semantic analysis (PLSA) model, where color and texture descriptors extracted from same image patches are jointly modeled with their conditional distributions. Then the proposed dual constraints: visual words importance and local image manifold are embedded into the joint-feature PLSA model simultaneously to obtain discriminative latent semantic topics. The visual word importance is proposed in our DJTM to guarantee that visual words with similar importance come from close latent topics while the local image manifold constraint enforces that images within the same category share similar latent topics. Finally, each image is characterized by distribution of latent semantic topics instead of low level features. Our proposed DJTM showed an excellent overall recognition accuracy 90.78%. Comprehensive comparison results demonstrate that our method outperforms existing multiple abnormalities classification methods for WCE images.

Hyperspectral image (HSI) classification is an active and important research task driven by many practical applications. To leverage deep learning models especially convolutional neural networks (CNNs) for HSI classification, this paper proposes a simple yet effective method to extract hierarchical deep spatial feature for HSI classification by exploring the power of off-the-shelf CNN models, without any additional retraining or fine-tuning on the target data set. To obtain better classification accuracy, we further propose a unified metric learning-based framework to alternately learn discriminative spectral-spatial features, which have better representation capability and train support vector machine (SVM) classifiers. To this end, we design a new objective function that explicitly embeds a metric learning regularization term into SVM training. The metric learning regularization term is used to learn a powerful spectral-spatial feature representation by fusing spectral feature and deep spatial feature, which has small intraclass scatter but big between class separation. By transforming HSI data into new spectral-spatial feature space through CNN and metric learning, we can pull the pixels from the same class closer, while pushing the different class pixels farther away. In the experiments, we comprehensively evaluate the proposed method on three commonly used HSI benchmark data sets. State-of-the-art results are achieved when compared with the existing HSI classification methods.

Remote sensing image scene classification is an active and challenging task driven by many applications. More recently, with the advances of deep learning models especially convolutional neural networks (CNNs), the performance of remote sensing image scene classification has been significantly improved due to the powerful feature representations learnt through CNNs. Although great success has been obtained so far, the problems of within-class diversity and between-class similarity are still two big challenges. To address these problems, in this paper, we propose a simple but effective method to learn discriminative CNNs (D-CNNs) to boost the performance of remote sensing image scene classification. Different from the traditional CNN models that minimize only the cross entropy loss, our proposed D-CNN models are trained by optimizing a new discriminative objective function. To this end, apart from minimizing the classification error, we also explicitly impose a metric learning regularization term on the CNN features. The metric learning regularization enforces the D-CNN models to be more discriminative so that, in the new D-CNN feature spaces, the images from the same scene class are mapped closely to each other and the images of different classes are mapped as farther apart as possible. In the experiments, we comprehensively evaluate the proposed method on three publicly available benchmark data sets using three off-the-shelf CNN models. Experimental results demonstrate that our proposed D-CNN methods outperform the existing baseline methods and achieve state-of-the-art results on all three data sets.

Image set classification has attracted much attention because of its broad applications. Despite the success made so far, the problems of intra-class diversity and inter-class similarity still remain two major challenges. To explore a possible solution to these challenges, this paper proposes a novel approach, termed duplex metric learning (DML), for image set classification. The proposed DML consists of two progressive metric learning stages with different objectives used for feature learning and image classification, respectively. The metric learning regularization is not only used to learn powerful feature representations but also well explored to train an effective classifier. At the first stage, we first train a discriminative stacked autoencoder (DSAE) by layer-wisely imposing a metric learning regularization term on the neurons in the hidden layers and meanwhile minimizing the reconstruction error to obtain new feature mappings in which similar samples are mapped closely to each other and dissimilar samples are mapped farther apart. At the second stage, we discriminatively train a classifier and simultaneously fine-tune the DSAE by optimizing a new objective function, which consists of a classification error term and a metric learning regularization term. Finally, two simple voting strategies are devised for image set classification based on the learnt classifier. In the experiments, we extensively evaluate the proposed framework for the tasks of face recognition, object recognition, and face verification on several commonly-used data sets and state-of-the-art results are achieved in comparison with existing methods.

The discriminability of Bag-of-Words representations can be increased via encoding the spatial relationship among virtual words on 3D shapes. However, this encoding task involves several issues, including arbitrary mesh resolutions, irregular vertex topology, orientation ambiguity on 3D surface, invariance to rigid and non-rigid shape transformations. To address these issues, a novel unsupervised spatial learning framework based on deep neural network, deep spatiality (DS), is proposed. Specifically, DS employs two novel components: spatial context extractor and deep context learner. Spatial context extractor extracts the spatial relationship among virtual words in a local region into a raw spatial representation. Along a consistent circular direction, a directed circular graph is constructed to encode relative positions between pairwise virtual words in each face ring into a relative spatial matrix. By decomposing each relative spatial matrix using SVD, the raw spatial representation is formed, from which deep context learner conducts unsupervised learning of global and local features. Deep context learner is a deep neural network with a novel model structure to adapt the proposed coupled softmax layer, which encodes not only the discriminative information among local regions but also the one among global shapes. Experimental results show that DS outperforms state-of-the-art methods.

Given a set of images that contain objects from a common category, object co-segmentation aims at automatically discovering and segmenting such common objects from each image. During the past few years, object co-segmentation has received great attention in the computer vision community. However, the existing approaches are usually designed with misleading assumptions, unscalable priors, or subjective computational models, which do not have sufficient robustness for dealing with complex and unconstrained real-world image contents. This paper proposes a novel two-stage co-segmentation framework, mainly for addressing the robustness issue. In the proposed framework, we first introduce the concept of union background and use it to improve the robustness for suppressing the image backgrounds contained by the given image groups. Then, we also weaken the requirement for the strong prior knowledge by using the background prior instead. This can improve the robustness when scaling up for the unconstrained image contents. Based on the weak background prior, we propose a novel MR-SGS model, i.e., manifold ranking with the self-learned graph structure, which can infer suitable graph structures in a data-driven manner rather than building the fixed graph structure relying on the subjective design. Such capacity is critical for further improving the robustness in inferring the foreground/background probability of each image pixel. Comprehensive experiments and comparisons with other state-of-the-art approaches can demonstrate the effectiveness of the proposed work.

Crowd counting is a challenging task, mainly due to the severe occlusions among dense crowds. This paper aims to take a broader view to address crowd counting from the perspective of semantic modeling. In essence, crowd counting is a task of pedestrian semantic analysis involving three key factors: pedestrians, heads, and their context structure. The information of different body parts is an important cue to help us judge whether there exists a person at a certain position. Existing methods usually perform crowd counting from the perspective of directly modeling the visual properties of either the whole body or the heads only, without explicitly capturing the composite body-part semantic structure information that is crucial for crowd counting. In our approach, we first formulate the key factors of crowd counting as semantic scene models. Then, we convert the crowd counting problem into a multi-task learning problem, such that the semantic scene models are turned into different sub-tasks. Finally, the deep convolutional neural networks are used to learn the sub-tasks in a unified scheme. Our approach encodes the semantic nature of crowd counting and provides a novel solution in terms of pedestrian semantic analysis. In experiments, our approach outperforms the state-of-the-art methods on four benchmark crowd counting data sets. The semantic structure information is demonstrated to be an effective cue in scene of crowd counting.

Traditional saliency models usually adopt hand-crafted image features and human-designed mechanisms to calculate local or global contrast. In this paper, we propose a novel computational saliency model, i.e., deep spatial contextual long-term recurrent convolutional network (DSCLRCN), to predict where people look in natural scenes. DSCLRCN first automatically learns saliency related local features on each image location in parallel. Then, in contrast with most other deep network based saliency models which infer saliency in local contexts, DSCLRCN can mimic the cortical lateral inhibition mechanisms in human visual system to incorporate global contexts to assess the saliency of each image location by leveraging the deep spatial long short-term memory (DSLSTM) model. Moreover, we also integrate scene context modulation in DSLSTM for saliency inference, leading to a novel deep spatial contextual LSTM (DSCLSTM) model. The whole network can be trained end-to-end and works efficiently when testing. Experimental results on two benchmark datasets show that DSCLRCN can achieve state-of-the-art performance on saliency detection. Furthermore, the proposed DSCLSTM model can significantly boost the saliency detection performance by incorporating both global spatial interconnections and scene context modulation, which may uncover novel inspirations for studies on them in computational saliency models.

Place-of-Interest (POI) summarization by aesthetics evaluation can recommend a set of POI images to the user and it is significant in image retrieval. In this paper, we propose a system that summarizes a collection of POI images regarding both aesthetics and diversity of the distribution of cameras. First, we generate visual albums by a coarse-to-fine POI clustering approach and then generate 3D models for each album by the collected images from social media. Second, based on the 3D to 2D projection relationship, we select candidate photos in terms of the proposed crowd source saliency model. Third, in order to improve the performance of aesthetic measurement model, we propose a crowd-sourced saliency detection approach by exploring the distribution of salient regions in the 3D model. Then, we measure the composition aesthetics of each image and we explore crowd source salient feature to yield saliency map, based on which, we propose an adaptive image adoption approach. Finally, we combine the diversity and the aesthetics to recommend aesthetic pictures. Experimental results show that the proposed POI summarization approach can return images with diverse camera distributions and aesthetics.

As the image sharing websites like Flickr become more and more popular, extensive scholars concentrate on tag-based image retrieval. It is one of the important ways to find images contributed by social users. In this research field, tag information and diverse visual features have been investigated. However, most existing methods use these visual features separately or sequentially. In this paper, we propose a global and local visual features fusion approach to learn the relevance of images by hypergraph approach. A hypergraph is constructed first by utilizing global, local visual features, and tag information. Then, we propose a pseudo-relevance feedback mechanism to obtain the pseudo-positive images. Finally, with the hypergraph and pseudo relevance feedback, we adopt the hypergraph learning algorithm to calculate the relevance score of each image to the query. Experimental results demonstrate the effectiveness of the proposed approach.

Weakly supervised video object segmentation (WSVOS) focuses on generating pixel-level object masks for videos only tagged with class labels, which is an essential yet challenging task. For WSVOS, the algorithm is just aware of rough category information rather than the concrete object size and location cues, besides it lacks reliable annotated exemplars to learn temporal evolution in the investigated videos. Basically, there are three challenging factors which may influence the performance of WSVOS: foreground object discovery in each frame, coarse object semantic consistency within each video, and fine-grained segmentation smoothness within neighbor frames. In this paper, we establish a semantic ranking and optical warping network (SROWN) to simultaneously solve these three challenges in a unified framework. For the first challenge, we apply the still image saliency detection method and discover the foreground object for each frame via a segmentation network. Due to the huge discrepancies between the image saliency and the video object segmentation, we step further and propose two subnetworks to solve the other two challenges. For the second one, we propose an attentive semantic ranking subnetwork to mine video-level tags, which can learn discriminative features for semantic ranking and lead to semantic consistent segmentation masks. For the third one, we propose an optical flow warping subnetwork to constrain fine-grained segmentation smoothness within neighbor frames, which can suppress the large deformation and thus obtain smooth object boundaries for adjacent frames. Experiments on two benchmark datasets, i.e., DAVIS dataset and YouTube-Objects dataset, demonstrate the effectiveness of the proposed approach for segmenting out video objects under weak supervision.

We propose a novel attention framework called attentive linear transformation (ALT). Instead of learning the spatial or channel-wise attention in existing models, ALT learns to attend to the high-dimensional transformation matrix from the image feature space to the context vector space. Thus ALT can learn various relevant feature abstractions, including spatial attention, channel-wise attention and visual dependence. Besides, we propose a soft threshold regression to predict the attention probabilities for local regions. Soft threshold regression preserves more useful visual information than popular softmax regression. Extensive experiments on the MS COCO and the Flickr30k datasets demonstrate the superiority of our model compared with other state-of-the-art methods.

Co-saliency detection is a newly emerging and rapidly growing research area in the computer vision community. As a novel branch of visual saliency, co-saliency detection refers to the discovery of common and salient foregrounds from two or more relevant images, and it can be widely used in many computer vision tasks. The existing co-saliency detection algorithms mainly consist of three components: extracting effective features to represent the image regions, exploring the informative cues or factors to characterize co-saliency, and designing effective computational frameworks to formulate co-saliency. Although numerous methods have been developed, the literature is still lacking a deep review and evaluation of co-saliency detection techniques. In this article, we aim at providing a comprehensive review of the fundamentals, challenges, and applications of co-saliency detection. Specifically, we provide an overview of some related computer vision works, review the history of co-saliency detection, summarize and categorize the major algorithms in this research area, discuss some open issues in this area, present the potential applications of co-saliency detection, and finally point out some unsolved challenges and promising future works. We expect this review to be beneficial to both fresh and senior researchers in this field and to give insights to researchers in other related areas regarding the utility of co-saliency detection algorithms.

Recently, intelligent vehicles catch much attention in both academia and industry. The vision-based moving object/vehicle detection and tracking are typically the core techniques for the event and activity analysis and the understanding of the dynamic driving environment in an intelligent vehicle. However, due to the complicated non-stationary environment, most existing vision-based motion tracking algorithms proposed for other simple conditions are not able to consistently track the objects. Therefore, in this paper, we propose a robust and long-term tracking method for intelligent vehicles, in which a set of classifiers are dynamically maintained and sampled for tackling varied challenges. In contrast to previous methods, to increase the diversity, a set of basic classifiers trained sequentially on different small data sets over time is dynamically maintained. The subsets of basic classifiers are independent with each other and can be specified to solve certain different sub-problems occurred in a non-stationary environment. Thus, for every challenge, an optimal classifier can be approximated in a subspace spanned by the selected competitive classifiers, which can address the current problem according to the distribution of the samples and recent performance. As a result, the tracker can efficiently address the various “concept drift” problems occurred together in a long video sequence. Due to the use of sparse weights for the competitive classifiers, the tracker can keep the balance between the efficiency and the performance. Experimental results show that the tracker yields competitive performance under various challenging environmental conditions and, especially, can overcome several challenges simultaneously.

Visual saliency detection has become an active research direction in recent years. A large number of saliency models, which can automatically locate objects of interest in images, have been developed. As these models take advantage of different kinds of prior assumptions, image features, and computational methodologies, they have their own strengths and weaknesses and may cope with only one or a few types of images well. Inspired by these facts, this paper proposes a novel salient object detection approach with the idea of inferring a superior model from a variety of previous imperfect saliency models via optimally leveraging the complementary information among them. The proposed approach mainly consists of three steps. First, a number of existing unsupervised saliency models are adopted to provide weak/imperfect saliency predictions for each region in the image. Then, a fusion strategy is used to fuse each image region's weak saliency predictions into a strong one by simultaneously considering the performance differences among various weak predictions and various characteristics of different image regions. Finally, a local spatial consistency constraint that ensures high similarity of the saliency labels for neighboring image regions with similar features is proposed to refine the results. Comprehensive experiments on five public benchmark datasets and comparisons with a number of state-of-the-art approaches can demonstrate the effectiveness of the proposed work.

Fast-growing scientific papers pose the problem of rapidly and accurately finding a list of reference papers for a given manuscript. Citation recommendation is an indispensable technique to overcome this obstacle. In this paper, we propose a citation recommendation approach via mutual reinforcement on a three-layered graph, in which each paper, author or venue is represented as a vertex in the paper layer, author layer, and venue layer, respectively. For personalized recommendation, we initiate the random walk separately for each query researcher. However, this has a high computational complexity due to the large graph size. To solve this problem, we apply a three-layered interactive clustering approach to cluster related vertices in the graph. Personalized citation recommendations are then made on the subgraph, generated by the clusters associated with each researcher's needs. When evaluated on the ACL anthology network, DBLP, and CiteSeer ML data sets, the performance of our proposed model-based citation recommendation approach is comparable with that of other state-of-the-art citation recommendation approaches. The results also demonstrate that the personalized recommendation approach is more effective than the nonpersonalized recommendation approach.

Eye movements in the case of freely viewing natural scenes are believed to be guided by local contrast, global contrast, and top-down visual factors. Although a lot of previous works have explored these three saliency cues for several years, there still exists much room for improvement on how to model them and integrate them effectively. This paper proposes a novel computation model to predict eye fixations, which adopts a multiresolution convolutional neural network (Mr-CNN) to infer these three types of saliency cues from raw image data simultaneously. The proposed Mr-CNN is trained directly from fixation and nonfixation pixels with multiresolution input image regions with different contexts. It utilizes image pixels as inputs and eye fixation points as labels. Then, both the local and global contrasts are learned by fusing information in multiple contexts. Meanwhile, various top-down factors are learned in higher layers. Finally, optimal combination of top-down factors and bottom-up contrasts can be learned to predict eye fixations. The proposed approach significantly outperforms the state-of-the-art methods on several publically available benchmark databases, demonstrating the superiority of Mr-CNN. We also apply our method to the RGB-D image saliency detection problem. Through learning saliency cues induced by depth and RGB information on pixel level jointly and their interactions, our model achieves better performance on predicting eye fixations in RGB-D images.

The amount of matrix data has increased rapidly nowadays. How to classify matrix data efficiently is an important issue. In this paper, by discovering the shortages of 2-D linear discriminant analysis and 2-D logistic regression, a novel 2-D framework named rank- 2-D multinomial logistic regression (2DMLR-RK) is proposed. The 2DMLR-RK is designed for a multiclass matrix classification problem. In the proposed framework, each category is modeled by a left projection matrix and a right projection matrix with rank . The left projection matrices capture the row information of matrix data, and the right projection matrices acquire the column information. We identify the parameter plays the role of balancing the capacity of learning and generalization of the 2DMLR-RK. In addition, we develop an effective framework for solving the proposed nonconvex optimization problem. The convergence, initialization, and computational complexity are discussed. Extensive experiments on various types of data sets are conducted. Comparing with 1-D methods, 2DMLR-RK not only achieves a better classification accuracy, but also costs less computation time. Comparing with other state-of-the-art 2-D methods, the 2DMLR-RK achieves a better performance for matrix data classification.

A large family of algorithms for unsupervised dimension reduction is based on both the local and global structures of the data. A fundamental step in these methods is to model the local geometrical structure of the data. However, the previous methods mainly ignore two facts in this step: 1) the dimensionality of the data is usually far larger than the number of local data, which is a typical ill-posed problem and 2) the data might be polluted by noise. These facts normally may lead to an inaccurate learned local structure and may degrade the final performance. In this paper, we propose a novel unsupervised dimension reduction method with the ability to address these problems effectively while also preserving the global information of the input data. Specifically, we first denoise the local data by preserving their principal components and we then apply a regularization term to the local modeling function to solve the illposed problem. Then, we use a linear regression model to capture the local geometrical structure, which is demonstrated to be insensitive to the parameters. Finally, we propose two criteria to simultaneously model both the local and the global information. Theoretical analyses for the relations between the proposed methods and some classical dimension-reduction methods are presented. The experimental results from various databases demonstrate the effectiveness of our methods.

With the development of the information technology, the amount of data, e.g. text, image and video, has been increased rapidly. Efficiently clustering those large scale data sets is a challenge. To address this problem, this paper proposes a novel co-clustering method named bilateral k-means algorithm (BKM) for fast co-clustering. Different from traditional k-means algorithms, the proposed method has two indicator matrices P and Q and a diagonal matrix S to be solved, which represent the cluster memberships of samples and features, and the co-cluster centres, respectively. Therefore, it could implement different clustering tasks on the samples and features simultaneously. We also introduce an effective approach to solve the proposed method, which involves less multiplication. The computational complexity is analyzed. Extensive experiments on various types of data sets are conducted. Compared with the state-of-the-art clustering methods, the proposed BKM not only has faster computational speed, but also achieves promising clustering results.

Clustering is a fundamental research topic in data mining. A balanced clustering result is often required in a variety of applications. Many existing clustering algorithms have good clustering performances, yet fail in producing balanced clusters. In this paper, we propose a novel and simple method for clustering, referred to as the Balanced Clustering with Least Square regression (BCLS), to minimize the least square linear regression, with a balance constraint to regularize the clustering model. In BCLS, the linear regression is applied to estimate the class-specific hyperplanes that partition each class of data from others, thus guiding the clustering of the data points into different clusters. A balance constraint is utilized to regularize the clustering, by minimizing which can help produce balanced clusters. In addition, we apply the method of augmented Lagrange multipliers (ALM) to help optimize the objective model. The experiments on seven real-world benchmarks demonstrate that our approach not only produces good clustering performance but also guarantees a balanced clustering result.

We introduce a novel supervised metric learning algorithm named parameter free large margin nearest neighbor (PFLMNN) which can be seen as an improvement of the classical large margin nearest neighbor (LMNN) algorithm. The contributions of our work consist of two aspects. First, our method discards the costterm which shrinks the distances between inquiry input and its k target neighbors (the k nearest neighbors with same labels as inquiry input) in LMNN, and only focuses on improving the action to push the imposters (the samples with different labels form the inquiry input) apart out of the neighborhood of inquiry. As a result, our method does not have the parameter needed to tune on the validating set, which makes it more convenient to use. Second, by leveraging the geometry information of the imposters, we construct a novel cost function to penalize the smalldistances between each inquiry and its imposters. Different from LMNN considering every imposter located in the neighborhood of each inquiry, our method only takes care of the nearest imposters. Because when the nearest imposter is pushed out of the neighborhood of its inquiry, other imposters would be all out. In this way, the constraints in our model are much less than that of LMNN, which makes our method much easier to find the optimal distance metric. Consequently, our method not only learns a better distance metric than LMNN, but also runs faster than LMNN. Extensive experiments on different data sets with various sizes and difficulties are conducted, and the results have shown that, compared with LMNN, PFLMNN achieves better classification results.

Recently, researchers have made great processes to build category-specific 3D shape models from 2D images with manual annotations consisting of class labels, keypoints, and ground truth figure-ground segmentations. However, the annotation of figure-ground segmentations is still labor-intensive and time-consuming. To further alleviate the burden of providing such manual annotations, we make the earliest effort to learn category-specific 3D shape models by only using weakly labeled 2D images. By revealing the underlying relationship between the tasks of common object segmentation and category-specific 3D shape reconstruction, we propose a novel framework to jointly solve these two problems along a cluster-level learning curriculum. Comprehensive experiments on the challenging PASCAL VOC benchmark demonstrate that the category-specific 3D shape models trained using our weakly supervised learning framework could, to some extent, approach the performance of the state-of-the-art methods using expensive manual segmentation annotations. In addition, the experiments also demonstrate the effectiveness of using 3D shape models for helping common object segmentation.

Object segmentation in weakly labelled videos is an interesting yet challenging task, which aims at learning to perform category-specific video object segmentation by only using video-level tags. Existing works in this research area might still have some limitations, e.g., lack of effective DNN-based learning frameworks, under-exploring the context information, and requiring to leverage the unstable negative video collection, which prevent them from obtaining more promising performance. To this end, we propose a novel self-paced fine-tuning network (SPFTN)-based framework, which could learn to explore the context information within the video frames and capture adequate object semantics without using the negative videos. To perform weakly supervised learning based on the deep neural network, we make the earliest effort to integrate the self-paced learning regime and the deep neural network into a unified and compatible framework, leading to the self-paced fine-tuning network. Comprehensive experiments on the large-scale YouTube-Objects and DAVIS datasets demonstrate that the proposed approach achieves superior performance as compared with other state-of-the-art methods as well as the baseline networks and models.

Spectral clustering has been widely used due to its simplicity for solving graph clustering problem in recent years. However, it suffers from the high computational cost as data grow in scale, and is limited by the performance of post-processing. To address these two problems simultaneously, in this paper, we propose a novel approach denoted by orthogonal and nonnegative graph reconstruction (ONGR) that scales linearly with the data size. For the relaxation of Normalized Cut, we add nonnegative constraint to the objective. Due to the nonnegativity, ONGR offers interpretability that the final cluster labels can be directly obtained without post-processing. Extensive experiments on clustering tasks demonstrate the effectiveness of the proposed method.

Mixture of regressions (MoR) is the well-established and effective approach to model discontinuous and heterogeneous data in regression problems. Existing MoR approaches assume smooth joint distribution for its good anlaytic properties. However, such assumption makes existing MoR very sensitive to intra-component outliers (the noisy training data residing in certain components) and the inter-component imbalance (the different amounts of training data in different components). In this paper, we make the earliest effort on Self-paced Learning (SPL) in MoR, i.e., Self-paced mixture of regressions (SPMoR) model. We propose a novel self-paced regularizer based on the Exclusive LASSO, which improves inter-component balance of training data. As a robust learning regime, SPL pursues confidence sample reasoning. To demonstrate the effectiveness of SPMoR, we conducted experiments on both the sythetic examples and real-world applications to age estimation and glucose estimation. The results show that SPMoR outperforms the state-of-the-arts methods.

Many graph based semi-supervised dimensionality reduction algorithms utilize the projection matrix to linearly map the data matrix from the original feature space to a lower dimensional representation. But the dimensionality after reduction is inevitably restricted to the number of classes, and the learned non-orthogonal projection matrix usually fails to preserve distances well and balance the weight on different projection direction. This paper proposes a novel dimensionality reduction method, called the semi-supervised orthogonal graph embedding with recursive projections (SOGE). We integrate the manifold smoothness and label fitness as well as the penalization of the linear mapping mismatch, and learn the orthogonal projection on the Stiefel manifold that empirically demonstrates better performance. Moreover, we recursively update the projection matrix in its orthocomplemented space to continuously learn more projection vectors, so as to better control the dimension of reduction. Comprehensive experiment on several benchmarks demonstrates the significant improvement over the existing methods.

In this paper, we propose a novel linear subspace learning algorithm called Flexible Orthogonal Neighborhood Preserving Embedding (FONPE), which is a linear approximation of Locally Linear Embedding (LLE) algorithm. Our novel objective function integrates two terms related to manifold smoothness and a flexible penalty defined on the projection fitness. Different from Neighborhood Preserving Embedding (NPE), we relax the hard constraint by modeling the mismatch between the approximate linear embedding and the original nonlinear embedding instead of enforcing them to be equal, which makes it better cope with the data sampled from a nonlinear manifold. Besides, instead of enforcing an orthogonality between the projected points, we enforce the mapping to be orthogonal. By using this method, FONPE tends to preserve distances and thus the overall geometry can be preserved. Unlike LLE, as FONPE has an explicit linear mapping between the input and the reduced spaces, it can handle novel testing data straightforwardly. Moreover, when the projection matrix in our model becomes an identity matrix, our model can be transformed to denoising LLE (DLLE). Compared with the standard LLE, we demonstrate that DLLE can handle data with noise better. Comprehensive experiments on several benchmark databases demonstrate the effectiveness of our algorithm.

Matrices are common forms of data that are encountered in a wide range of real applications. How to classify this kind of data is an important research topic. In this paper, we propose a novel distance metric learning method named two dimensional large margin nearest neighbor (2DLMNNN), for improving the performance of k nearest neighbor (KNN) classifier in matrix classification. In the proposed method, left and right projection matrices are employed to define the matrix-based Mahalanobis distance, which is used to construct the objective aimed at separating points in different classes by a large margin. The parameters in those two projection matrices are much less than that in its vector-based counterpart, thus our method reduces the risks of overfitting. We also introduce a framework for solving the proposed 2DLMNN. The convergence behavior, initialization, and parameter determination are also analyzed. Compared with vector-based methods, 2DLMNN performs better for matrix data classification. Promising experimental results on several data sets are provided to demonstrate the effectiveness of our method.

Many previous graph-based methods perform dimensionality reduction on a pre-defined graph. However, due to the noise and redundant information in the original data, the pre-defined graph has no clear structure and may not be appropriate for the subsequent task. To overcome the drawbacks, in this paper, we propose a novel approach called linear manifold regularization with adaptive graph (LMRAG) for semi-supervised dimensionality reduction. LMRAG directly incorporates the graph construction into the objective function, thus the projection matrix and the optimal graph can be simultaneously optimized. Due to the structure constraint, the learned graph is sparse and has clear structure. Extensive experiments on several benchmark datasets demonstrate the effectiveness of the proposed method.

More and more multi-view data which can capture rich information from heterogeneous features are widely used in real world applications. How to integrate different types of features, and how to learn low dimensional and discriminative information from high dimensional data are two main challenges. To address these challenges, this paper proposes a novel multi-view feature learning framework, which is regularized by discriminative information and obtains a feature learning model that contains multiple discriminative feature weighting matrices for different views, and then yields multiple low dimensional features used for subsequent multi-view clustering. To optimize the formulated objective function, we transform the proposed framework into a trace optimization problem which obtains the global solution in a closed form. Experimental evaluations on four widely used datasets and comparisons with a number of state-of-the-art multi-view clustering algorithms demonstrate the superiority of the proposed work.

In data mining, we often encounter high dimensional and noisy features, which may not only increase the load of computational resources but also result in the problem of model overfitting. Feature selection is often adopted to address this issue. In this paper, we propose a novel feature selection method based on multi-class SVM, which introduces the scaling factor with a flexible parameter to renewedly adjust the distribution of feature weights and select the most discriminative features. Concretely, the proposed method designs a scaling factor with p/2 power to control the distribution of weights adaptively and search optimal sparsity of weighting matrix. In addition, to solve the proposed model, we provide an alternative and iterative optimization method. It not only makes solutions of weighting matrix and scaling factor independently, but also provides a better way to address the problem of solving L2,0-norm. Comprehensive experiments are conducted on six datasets to demonstrate that this work can obtain better performance compared with a number of existing state-of-the-art multi-class feature selection methods.

As an interesting and emerging topic, co-saliency detection aims at simultaneously extracting common salient objects from a group of images. On one hand, traditional co-saliency detection approaches rely heavily on human knowledge for designing hand-crafted metrics to possibly reflect the faithful properties of the co-salient regions. Such strategies, however, always suffer from poor generalization capability to flexibly adapt various scenarios in real applications. On the other hand, most current methods pursue co-saliency detection in unsupervised fashions. This, however, tends to weaken their performance in real complex scenarios because they are lack of robust learning mechanism to make full use of the weak labels of each image. To alleviate these two problems, this paper proposes a new SP-MIL framework for co-saliency detection, which integrates both multiple instance learning (MIL) and self-paced learning (SPL) into a unified learning framework. Specifically, for the first problem, we formulate the co-saliency detection problem as a MIL paradigm to learn the discriminative classifiers to detect the co-saliency object in the "instance-level". The formulated MIL component facilitates our method capable of automatically producing the proper metrics to measure the intra-image contrast and the inter-image consistency for detecting co-saliency in a purely self-learning way. For the second problem, the embedded SPL paradigm is able to alleviate the data ambiguity under the weak supervision of co-saliency detection and guide a robust learning manner in complex scenarios. Experiments on benchmark datasets together with multiple extended computer vision applications demonstrate the superiority of the proposed framework beyond the state-of-the-arts.

Remote sensing image scene classification plays an important role in a wide range of applications and hence has been receiving remarkable attention. During the past years, significant efforts have been made to develop various data sets or present a variety of approaches for scene classification from remote sensing images. However, a systematic review of the literature concerning data sets and methods for scene classification is still lacking. In addition, almost all existing data sets have a number of limitations, including the small scale of scene classes and the image numbers, the lack of image variations and diversity, and the saturation of accuracy. These limitations severely limit the development of new approaches especially deep learning-based methods. This paper first provides a comprehensive review of the recent progress. Then, we propose a large-scale data set, termed “NWPU-RESISC45,” which is a publicly available benchmark for REmote Sensing Image Scene Classification (RESISC), created by Northwestern Polytechnical University (NWPU). This data set contains 31 500 images, covering 45 scene classes with 700 images in each class. The proposed NWPU-RESISC45 1) is large-scale on the scene classes and the total image number; 2) holds big variations in translation, spatial resolution, viewpoint, object pose, illumination, background, and occlusion; and 3) has high within-class diversity and between-class similarity. The creation of this data set will enable the community to develop and evaluate various data-driven algorithms. Finally, several representative methods are evaluated using the proposed data set, and the results are reported as a useful baseline for future research.

With fast development of information engineering and social network, people's locations can be conveniently sensed by spatial technology, such as global positioning systems (GPS), base stations, Wi-Fi access points and even from the appearances of the photos they have taken. The social networks and the online shopping platforms have been gathering billions of users, who share a large amount of images taken in places they live in and visit. We can leverage the social networks to express our opinions about the services and places of interest (POIs). The interactions among users, and user and POIs or services generate big social media data, which have rich information for user, location, and service cognition. Many real-time network applications rely heavily on the accurate social users' locations. How to sense the locations from multisource social media data is very important and challenging. Thus, in this paper, we give a systematic review of the works that combine social media and spatial technology for POI cognition and image localization.

The advent of low-cost depth cameras, such as the Microsoft Kinect in the consumer market, has made many indoor applications and games based on motion tracking available to the everyday user. However, it is a large challenge to track human motion via such a camera because of its low-quality images, missing depth values, and noise. In this paper, we propose a novel human motion capture method based on a cooperative structure of multiple low-cost RGBD cameras, which can effectively avoid these problems. This structure can also manage the problem of body occlusions that appears when a single camera is used. Moreover, the whole process does not require training data, which makes this approach easily deployed and reduces operation time. We use the color image, depth image, and point cloud acquired in each view as the data source, and an initial pose is extracted in our optimization framework by aligning multiple point clouds from different cameras. The pose is dynamically updated by combining a filtering approach with a Markov model to estimate new poses in video streams. To verify the efficiency and robustness of our approach, we capture a wide variety of human actions via three cameras in indoor scenes and compare the tracking results of the proposed method to those of the current state-of-the-art methods. Moreover, our system is tested on more complex situations, in which multiple humans move within a scene, possibly occluding each other to some extent. The actions of multiple humans are tracked simultaneously, which would assist group behavior analysis.

Full human body shape scans provide valuable data for a variety of applications including anthropometric surveying, clothing design, human-factors engineering, health, and entertainment. However, the high price, large volume, and difficulty of operating professional 3-D scanners preclude their use in home entertainment. Recently, portable low-cost red green blue-depth cameras such as the Kinect have become popular for computer vision tasks. However, the infrared mechanism of this type of camera leads to noisy and incomplete depth images. We construct a stereo full-body scanning environment composed of multiple depth cameras and propose a novel registration algorithm. Our algorithm determines a segment constrained correspondence for two neighboring views, integrating them using rigid transformation. Furthermore, it aligns all of the views based on uniform error distribution. The generated 3-D mesh model is typically sparse, noisy, and even with holes, which makes it lose surface details. To address this, we introduce a geometric and topological fitting prior in the form of a professionally designed high-resolution template model. We formulate a template deformation optimization problem to fit the high-resolution model to the low-quality scan. Its solution overcomes the obstacles posed by different poses, varying body details, and surface noise. The entire process is free of body and template markers, fully automatic, and achieves satisfactory reconstruction results.

Fault detection and isolation (FDI) is very difficult for electronics-rich analog systems due to its sophisticated mechanism and variable operational conditions. Traditionally, FDI in such systems is done through the monitoring of deviation of output signals in voltage or current at system level, which commonly arises from the degradation of one or more critical components. Therefore, FDI can be transformed to a multiclass classification task given the extracted features of the output signals in voltage or current of the circuit. Traditional feature extraction on the circuit output is mostly based on time-domain, frequency-domain, or time-frequency signal processing, which collapse high-dimensional raw signals into a lower dimensional feature set. Such low-dimensional feature set usually suffers from information loss so as to affect the accuracy of the later fault diagnosis. In order to retain as much information as possible, deep learning is proposed which employs a hierarchical structure to capture the different levels of semantic representations of the signals. In this paper, a novel fault diagnostic application of Gaussian-Bernoulli deep belief network (GB-DBN) for electronics-rich analog systems is developed which can more effectively capture the high-order semantic features within the raw output signals. The novel fault diagnosis is validated experimentally on two typical analog filter circuits. Experimental results show the fault diagnosis based on GB-DBN is with superior diagnostic performance than the traditional feature extraction methods.

Highly discriminative 3D shape representations can be formed by encoding the spatial relationship among virtual words into the Bag of Words (BoW) method. To achieve this challenging task, several unresolved issues in the encoding procedure must be overcome for 3D shapes, including: 1) arbitrary mesh resolution; 2) irregular vertex topology; 3) orientation ambiguity on the 3D surface; and 4) invariance to rigid and non-rigid shape transformations. In this paper, a novel spatially enhanced 3D shape representation called bag of spatial context correlations (BoSCCs) is proposed to address all these issues. Adopting a novel local perspective, BoSCC is able to describe a 3D shape by an occurrence frequency histogram of spatial context correlation patterns, which makes BoSCC become more compact and discriminative than previous global perspective-based methods. Specifically, the spatial context correlation is proposed to simultaneously encode the geometric and spatial information of a 3D local region by the correlation among spatial contexts of vertices in that region, which effectively resolves the aforementioned issues. The spatial context of each vertex is modeled by Markov chains in a multi-scale manner, which thoroughly captures the spatial relationship by the transition probabilities of intra-virtual words and the ones of inter-virtual words. The high discriminability and compactness of BoSCC are effective for classification and retrieval, especially in the scenarios of limited samples and partial shape retrieval. Experimental results show that BoSCC outperforms the state-of-the-art spatially enhanced BoW methods in three common applications: global shape retrieval, shape classification, and partial shape retrieval.

Recent years, more and more multi-view data are widely used in many real world applications. This kind of data (such as image data) are high dimensional and obtained from different feature extractors, which represents distinct perspectives of the data. How to cluster such data efficiently is a challenge. In this paper, we propose a novel multi-view clustering framework, called Re-weighted Discriminatively Embedded KMeans (RDEKM), for this task. The proposed method is a multiview least-absolute residual model which induces robustness to efficiently mitigates the influence of outliers and realizes dimension reduction during multi-view clustering. Specifically, the proposed model is an unsupervised optimization scheme which utilizes Iterative Re-weighted Least Squares to solve leastabsolute residual and adaptively controls the distribution of multiple weights in a re-weighted manner only based on its own low-dimensional subspaces and a common clustering indicator matrix. Furthermore, theoretical analysis (including optimality and convergence analysis) and the optimization algorithm are also presented. Compared to several state-of-the-art multi-view clustering methods, the proposed method substantially improves the accuracy of the clustering results on widely used benchmark datasets, which demonstrates the superiority of the proposed work.

With the goal of discovering the common and salient objects from the given image group, co-saliency detection has received tremendous research interest in recent years. However, as most of the existing co-saliency detection methods are performed based on the assumption that all the images in the given image group should contain co-salient objects in only one category, they can hardly be applied in practice, particularly for the large-scale image set obtained from the Internet. To address this problem, this paper revisits the co-saliency detection task and advances its development into a new phase, where the problem setting is generalized to allow the image group to contain objects in arbitrary number of categories and the algorithms need to simultaneously detect multi-class co-salient objects from such complex data. To solve this new challenge, we decompose it into two sub-problems, i.e., how to identify subgroups of relevant images and how to discover relevant co-salient objects from each subgroup, and propose a novel co-saliency detection framework to correspondingly address the two sub-problems via two-stage multi-view spectral rotation co-clustering. Comprehensive experiments on two publically available benchmarks demonstrate the effectiveness of the proposed approach. Notably, it can even outperform the state-of-the-art co-saliency detection methods, which are performed based on the image subgroups carefully separated by the human labor.

The task of feature selection is to find the most representative features from the original high-dimensional data. Because of the absence of the information of class labels, selecting the appropriate features in unsupervised learning scenarios is much harder than that in supervised scenarios. In this paper, we investigate the potential of locally linear embedding (LLE), which is a popular manifold learning method, in feature selection task. It is straightforward to apply the idea of LLE to the graph-preserving feature selection framework. However, we find that this straightforward application suffers from some problems. For example, it fails when the elements in the feature are all equal; it does not enjoy the property of scaling invariance and cannot capture the change of the graph efficiently. To solve these problems, we propose a new filter-based feature selection method based on LLE in this paper, which is named as LLE score. The proposed criterion measures the difference between the local structure of each feature and that of the original data. Our experiments of classification task on two face image data sets, an object image data set, and a handwriting digits data set show that LLE score outperforms state-of-the-art methods, including data variance, Laplacian score, and sparsity score.

Recent progresses in multimedia event detection have enabled us to find videos about a predefined event from a large-scale video collection. Research towards more intrinsic unsupervised video understanding is an interesting but understudied field. Specifically, given a collection of videos sharing a common event of interest, the goal is to discover the salient fragments, i.e., the curt video fragments that can concisely portray the underlying event of interest, from each video. To explore this novel direction, this paper proposes an unsupervised event saliency revealing framework. It first extracts features from multiple modalities to represent each shot in the given video collection. Then, these shots are clustered to build the cluster-level event saliency revealing framework, which explores useful information cues (i.e., the intra-cluster prior, inter-cluster discriminability, and inter-cluster smoothness) by a concise optimization model. Compared with the existing methods, our approach could highlight the intrinsic stimulus of the unseen event within a video in an unsupervised fashion. Thus, it could potentially benefit to a wide range of multimedia tasks like video browsing, understanding, and search. To quantitatively verify the proposed method, we systematically compare the method to a number of baseline methods on the TRECVID benchmarks. Experimental results have demonstrated its effectiveness and efficiency.

Discriminative features of 3-D meshes are significant to many 3-D shape analysis tasks. However, handcrafted descriptors and traditional unsupervised 3-D feature learning methods suffer from several significant weaknesses: 1) the extensive human intervention is involved; 2) the local and global structure information of 3-D meshes cannot be preserved, which is in fact an important source of discriminability; 3) the irregular vertex topology and arbitrary resolution of 3-D meshes do not allow the direct application of the popular deep learning models; 4) the orientation is ambiguous on the mesh surface; and 5) the effect of rigid and nonrigid transformations on 3-D meshes cannot be eliminated. As a remedy, we propose a deep learning model with a novel irregular model structure, called mesh convolutional restricted Boltzmann machines (MCRBMs). MCRBM aims to simultaneously learn structure-preserving local and global features from a novel raw representation, local function energy distribution. In addition, multiple MCRBMs can be stacked into a deeper model, called mesh convolutional deep belief networks (MCDBNs). MCDBN employs a novel local structure preserving convolution (LSPC) strategy to convolve the geometry and the local structure learned by the lower MCRBM to the upper MCRBM. LSPC facilitates resolving the challenging issue of the orientation ambiguity on the mesh surface in MCDBN. Experiments using the proposed MCRBM and MCDBN were conducted on three common aspects: global shape retrieval, partial shape retrieval, and shape correspondence. Results show that the features learned by the proposed methods outperform the other state-of-the-art 3-D shape features.

Thanks to the powerful feature representations obtained through deep convolutional neural network (CNN), the performance of object detection has recently been substantially boosted. Despite the remarkable success, the problems of object rotation, within-class variability, and between-class similarity remain several major challenges. To address these problems, this paper proposes a novel and effective method to learn a rotation-invariant and Fisher discriminative CNN (RIFD-CNN) model. This is achieved by introducing and learning a rotation-invariant layer and a Fisher discriminative layer, respectively, on the basis of the existing high-capacity CNN architectures. Specifically, the rotation-invariant layer is trained by imposing an explicit regularization constraint on the objective function that enforces invariance on the CNN features before and after rotating. The Fisher discriminative layer is trained by imposing the Fisher discrimination criterion on the CNN features so that they have small within-class scatter but large between-class separation. In the experiments, we comprehensively evaluate the proposed method for object detection task on a public available aerial image dataset and the PASCAL VOC 2007 dataset. State-of-the-art results are achieved compared with the existing baseline methods.

Traditional salient object detection models often use hand-crafted features to formulate contrast and various prior knowledge, and then combine them artificially. In this work, we propose a novel end-to-end deep hierarchical saliency network (DHSNet) based on convolutional neural networks for detecting salient objects. DHSNet first makes a coarse global prediction by automatically learning various global structured saliency cues, including global contrast, objectness, compactness, and their optimal combination. Then a novel hierarchical recurrent convolutional neural network (HRCNN) is adopted to further hierarchically and progressively refine the details of saliency maps step by step via integrating local context information. The whole architecture works in a global to local and coarse to fine manner. DHSNet is directly trained using whole images and corresponding ground truth saliency masks. When testing, saliency maps can be generated by directly and efficiently feed forwarding testing images through the network, without relying on any other techniques. Evaluations on four benchmark datasets and comparisons with other 11 state-of-the-art algorithms demonstrate that DHSNet not only shows its significant superiority in terms of performance, but also achieves a real-time speed of 23 FPS on modern GPUs.

Aiming at automatically discovering the common objects contained in a set of relevant images and segmenting them as foreground simultaneously, object co-segmentation has become an active research topic in recent years. Although a number of approaches have been proposed to address this problem, many of them are designed with the misleading assumption, unscalable prior, or low flexibility and thus still suffer from certain limitations, which reduces their capability in the real-world scenarios. To alleviate these limitations, we propose a novel two-stage co-segmentation framework, which introduces the weak background prior to establish a globally close-loop graph to represent the common object and union background separately. Then a novel graph optimized-flexible manifold ranking algorithm is proposed to flexibly optimize the graph connection and node labels to co-segment the common objects. Experiments on three image datasets demonstrate that our method outperforms other state-of-the-art methods.

In real world applications, more and more data, for example, image/video data, are high dimensional and repre-sented by multiple views which describe different perspectives of the data. Efficiently clustering such data is a challenge. To address this problem, this paper proposes a novel multi-view clustering method called Discriminatively Embedded K-Means (DEKM), which embeds the synchronous learning of multiple discriminative subspaces into multi-view K-Means clustering to construct a unified framework, and adaptively control the intercoordinations between these subspaces simultaneously. In this framework, we firstly design a weighted multi-view Linear Discriminant Analysis (LDA), and then develop an unsupervised optimization scheme to alternatively learn the common clustering indicator, multiple discriminative subspaces and weights for heterogeneous features with convergence. Comprehensive evaluations on three benchmark datasets and comparisons with several state-of-the-art multi-view clustering algorithms demonstrate the superiority of the proposed work.

With the growing number of audio excerpts through various media and distribution channels, advanced audio analysis approaches have received significant interest in the multimedia field. However, current audio analysis approaches are still far from satisfactory due to the semantic gaps between the low-level acoustic features and high-level semantics perceived by human brain. In order to alleviate the problem, this paper propose a novel computational framework to bridge acoustic features with high-level semantic features derived from functional magnetic resonance imaging (fMRI) signals which record the brain's response during free listening to music/speech excerpts, and to explore the brain auditory network composition of acoustic features for different types of music/speech excerpts. Specifically, we identify meaningful brain networks and corresponding brain activities representing high-level semantic features via a novel group-wise sparse representation of whole brain fMRI signals. Then we associate the brain activities with specific low-level acoustic features and analyze the auditory network composition of acoustic features for different types of music/speech excerpts. Experimental results demonstrate that multiple acoustic features are involved in the brain auditory networks during free listening to music/speech excerpts. Meanwhile, there is considerable variability of auditory network composition of acoustic features for different types of music/speech. Our results provide new insights of how to narrow the semantic gaps in audio content analysis.

Scene classification of high resolution remote sensing images plays an important role for a wide range of applications. While significant efforts have been made in developing various methods for scene classification, most of them are based on handcrafted or shallow learning-based features. In this paper, we investigate the use of deep convolutional neural network (CNN) for scene classification. To this end, we first adopt two simple and effective strategies to extract CNN features: (1) using pre-trained CNN models as universal feature extractors, and (2) domain-specifically fine-tuning pre-trained CNN models on our scene classification dataset. Then, scene classification is carried out by using simple classifiers such as linear support vector machine (SVM). In our work, three off-the-shelf CNN models including AlexNet [1], VGGNet [2], and GoogleNet [3] are investigated. Comprehensive evaluations on a publicly available 21 classes land use dataset and comparisons with several state-of-the-art approaches demonstrate that deep CNN features are effective for scene classification of high resolution remote sensing images.

Automatic semantic annotation of high-resolution optical satellite images is a task to assign one or several predefined semantic concepts to an image according to its content. The fundamental challenge arises from the difficulty of characterizing complex and ambiguous contents of the satellite images. To address this challenge, a diversity constrained joint multi-feature learning method is proposed to learn robust feature representations for annotating satellite images. The key motivation of our method is to make full use of the complementarity diversity information among the heterogeneous features in the learning process. Comprehensive experiments on an annotation dataset demonstrate the superiority and effectiveness of our method compared with baseline multi-feature learning method.

Nowadays, there are many movie trailers publicly available on social media website such as YouTube, and many thousands of users have independently indicated whether they like or dislike those trailers. Although it is understandable that there are multiple factors that could influence viewers' like or dislike of the trailer, we aim to address a preference question in this work: Can subjective multimedia features be developed to predict the viewer's preference presented by like (by thumbs-up) or dislike (by thumbs-down) during and after watching movie trailers? We designed and implemented a computational framework that is composed of low-level multimedia feature extraction, feature screening and selection, and classification, and applied it to a collection of 725 movie trailers. Experimental results demonstrated that, among dozens of multimedia features, the single low-level multimedia feature of shot length variance is highly predictive of a viewer's “like/dislike” for a large portion of movie trailers. We interpret these findings such that variable shot lengths in a trailer tend to produce a rhythm that is likely to stimulate a viewer's positive preference. This conclusion was also proved by the repeatability experiments results using another 600 trailer videos and it was further interpreted by viewers'eye-tracking data.

Saliency detection models aiming to quantitatively predict human eye-attended locations in the visual field have been receiving increasing research interest in recent years. Unlike traditional methods that rely on hand-designed features and contrast inference mechanisms, this paper proposes a novel framework to learn saliency detection models from raw image data using deep networks. The proposed framework mainly consists of two learning stages. At the first learning stage, we develop a stacked denoising autoencoder (SDAE) model to learn robust, representative features from raw image data under an unsupervised manner. The second learning stage aims to jointly learn optimal mechanisms to capture the intrinsic mutual patterns as the feature contrast and to integrate them for final saliency prediction. Given the input of pairs of a center patch and its surrounding patches represented by the features learned at the first stage, a SDAE network is trained under the supervision of eye fixation labels, which achieves both contrast inference and contrast integration simultaneously. Experiments on three publically available eye tracking benchmarks and the comparisons with 16 state-of-the-art approaches demonstrate the effectiveness of the proposed framework.

Object detection in very high resolution optical remote sensing images is a fundamental problem faced for remote sensing image analysis. Due to the advances of powerful feature representations, machine-learning-based object detection is receiving increasing attention. Although numerous feature representations exist, most of them are handcrafted or shallow-learning-based features. As the object detection task becomes more challenging, their description capability becomes limited or even impoverished. More recently, deep learning algorithms, especially convolutional neural networks (CNNs), have shown their much stronger feature representation power in computer vision. Despite the progress made in nature scene images, it is problematic to directly use the CNN feature for object detection in optical remote sensing images because it is difficult to effectively deal with the problem of object rotation variations. To address this problem, this paper proposes a novel and effective approach to learn a rotation-invariant CNN (RICNN) model for advancing the performance of object detection, which is achieved by introducing and learning a new rotation-invariant layer on the basis of the existing CNN architectures. However, different from the training of traditional CNN models that only optimizes the multinomial logistic regression objective, our RICNN model is trained by optimizing a new objective function via imposing a regularization constraint, which explicitly enforces the feature representations of the training samples before and after rotating to be mapped close to each other, hence achieving rotation invariance. To facilitate training, we first train the rotation-invariant layer and then domain-specifically fine-tune the whole RICNN network to further boost the performance. Comprehensive evaluations on a publicly available ten-class object detection data set demonstrate the effectiveness of the proposed method.

In this paper, we focus on tackling the problem of automatic semantic annotation of high resolution (HR) optical satellite images, which aims to assign one or several predefined semantic concepts to an image according to its content. The main challenges arise from the difficulty of characterizing complex and ambiguous contents of the satellite images and the high human labor cost caused by preparing a large amount of training examples with high-quality pixel-level labels in fully supervised annotation methods. To address these challenges, we propose a unified annotation framework by combining discriminative high-level feature learning and weakly supervised feature transferring. Specifically, an efficient stacked discriminative sparse autoencoder (SDSAE) is first proposed to learn high-level features on an auxiliary satellite image data set for the land-use classification task. Inspired by the motivation that the encoder of the prelearned SDSAE can be regarded as a generic high-level feature extractor for HR optical satellite images, we then transfer the learned high-level features to semantic annotation. To compensate the difference between the auxiliary data set and the annotation data set, the transferred high-level features are further fine-tuned in a weakly supervised scheme by using the tile-level annotated training data. Finally, the fine-tuning process is formulated as an ultimate optimization problem, which can be solved efficiently with our proposed alternate iterative optimization method. Comprehensive experiments on a publicly available land-use classification data set and an annotation data set demonstrate the superiority of our SDSAE-based high-level feature learning method and the effectiveness of our weakly supervised semantic annotation framework compared with state-of-the-art fully supervised annotation methods.

Extracting local features from 3D shapes is an important and challenging task that usually requires carefully designed 3D shape descriptors. However, these descriptors are hand-crafted and require intensive human intervention with prior knowledge. To tackle this issue, we propose a novel deep learning model, namely circle convolutional restricted Boltzmann machine (CCRBM), for unsupervised 3D local feature learning. CCRBM is specially designed to learn from raw 3D representations. It effectively overcomes obstacles such as irregular vertex topology, orientation ambiguity on the 3D surface, and rigid or slightly non-rigid transformation invariance in the hierarchical learning of 3D data that cannot be resolved by the existing deep learning models. Specifically, by introducing the novel circle convolution, CCRBM holds a novel ring-like multi-layer structure to learn 3D local features in a structure preserving manner. Circle convolution convolves across 3D local regions via rotating a novel circular sector convolution window in a consistent circular direction. In the process of circle convolution, extra points are sampled in each 3D local region and projected onto the tangent plane of the center of the region. In this way, the projection distances in each sector window are employed to constitute a novel local raw 3D representation called projection distance distribution (PDD). In addition, to eliminate the initial location ambiguity of a sector window, the Fourier transform modulus is used to transform the PDD into the Fourier domain, which is then conveyed to CCRBM. Experiments using the learned local features are conducted on three aspects: global shape retrieval, partial shape retrieval, and shape correspondence. The experimental results show that the learned local features outperform other state-of-the-art 3D shape descriptors.

The paper presents a sketch-based image retrieval algorithm. One of the main challenges in sketch-based image retrieval (SBIR) is to measure the similarity between a sketch and an image. To tackle this problem, we propose an SBIR-based approach by salient contour reinforcement. In our approach, we divide the image contour into two types. The first is the global contour map. The second, called the salient contour map, is helpful to find out the object in images similar to the query. In addition, based on the two contour maps, we propose a new descriptor, namely an angular radial orientation partitioning (AROP) feature. It fully utilizes the edge pixels' orientation information in contour maps to identify the spatial relationships. Our AROP feature based on the two candidate contour maps is both efficient and effective to discover false matches of local features between sketches and images, and can greatly improve the retrieval performance. The application of the retrieval system based on this algorithm is established. The experiments on the image dataset with 0.3 million images show the effectiveness of the proposed method and comparisons with other algorithms are also given. Compared to baseline performance, the proposed method achieves 10% higher precision in top 5.

Vision and multimedia communities have long attempted to enable computers to understand image or video content in a manner analogous to humans. Humans’ comprehension to an image or a video clip often depends on the objects that draw their attention. As a result, one fundamental and open problem is to automatically infer the attention attracting or interesting areas in an image or a video sequence. Recently, a large number of researchers explore visual saliency models to address this problem. The study on visual saliency models is originally motivated by simulating humans’ bottom-up visual attention and it is mainly based on the biological evidence that humans’ visual attention is automatically attracted by highly salient features in the visual scene, which are discriminative with respect to the surrounding environment.

As an interesting and emerging topic, cosaliency detection aims at simultaneously extracting common salient objects in multiple related images. It differs from the conventional saliency detection paradigm in which saliency detection for each image is determined one by one independently without taking advantage of the homogeneity in the data pool of multiple related images. In this paper, we propose a novel cosaliency detection approach using deep learning models. Two new concepts, called intrasaliency prior transfer and deep intersaliency mining, are introduced and explored in the proposed work. For the intrasaliency prior transfer, we build a stacked denoising autoencoder (SDAE) to learn the saliency prior knowledge from auxiliary annotated data sets and then transfer the learned knowledge to estimate the intrasaliency for each image in cosaliency data sets. For the deep intersaliency mining, we formulate it by using the deep reconstruction residual obtained in the highest hidden layer of a self-trained SDAE. The obtained deep intersaliency can extract more intrinsic and general hidden patterns to discover the homogeneity of cosalient objects in terms of some higher level concepts. Finally, the cosaliency maps are generated by weighted integration of the proposed intrasaliency prior, deep intersaliency, and traditional shallow intersaliency. Comprehensive experiments over diverse publicly available benchmark data sets demonstrate consistent performance gains of the proposed method over the state-of-the-art cosaliency detection methods.

Implicit emotion tagging is a central theme in the area of affective computing. To this end, Several physiological signals acquired from subjects can be employed, for example, electroencephalography (EEG) and functional magnetic resonance imaging (fMRI) from brain, electrocardiography (ECG) from cardiac activities, and other peripheral physiological signals, such as galvanic skin resistance, electromyogram (EMG), blood volume pressure etc. Brain is regarded as the place where emotional activities evoke. Determining affective states by observing brain activities directly is of therefore great interest. There are several published works that use EEG signals to identify affective states in different aspects with various stimuli, e.s., images, musics and videos. In this paper, we propose to adopt EEG connectivity between electrodes to identify subjects' affective levels in both valence and arousal space during video stimuli presentation. Three catagories of connectivity are adopted in magnitude and phase domains. One open accessed affective database, DEAP, is used as benchmark. We will show that with the proposed connectivity-based representation, the accuracy of affective levels identification tasks are higher than the same tasks in existing works based on same database.

Part model-based methods have been successfully applied to object detection and scene classification and have achieved state-of-the-art results. More recently the “sparselets” work [1-3] were introduced to serve as a universal set of shared basis learned from a large number of part detectors, resulting in notable speedup. Inspired by this framework, in this paper, we propose a novel scheme to train more effective sparselets with a coarse-to-fine framework. Specifically, we first train coarse sparselets to exploit the redundancy existing among part detectors by using an unsupervised single-hidden-layer auto-encoder. Then, we simultaneously train fine sparselets and activation vectors using a supervised single-hidden-layer neural network, in which sparselets training and discriminative activation vectors learning are jointly embedded into a unified framework. In order to adequately explore the discriminative information hidden in the part detectors and to achieve sparsity, we propose to optimize a new discriminative objective function by imposing L0-norm sparsity constraint on the activation vectors. By using the proposed framework, promising results for multi-class object detection and scene classification are achieved on PASCAL VOC 2007, MIT Scene-67, and UC Merced Land Use datasets, compared with the existing sparselets baseline methods.

It is believed that eye movements in free-viewing of natural scenes are directed by both bottom-up visual saliency and top-down visual factors. In this paper, we propose a novel computational framework to simultaneously learn these two types of visual features from raw image data using a multiresolution convolutional neural network (Mr-CNN) for predicting eye fixations. The Mr-CNN is directly trained from image regions centered on fixation and non-fixation locations over multiple resolutions, using raw image pixels as inputs and eye fixation attributes as labels. Diverse top-down visual features can be learned in higher layers. Meanwhile bottom-up visual saliency can also be inferred via combining information over multiple resolutions. Finally, optimal integration of bottom-up and top-down cues can be learned in the last logistic regression layer to predict eye fixations. The proposed approach achieves state-of-the-art results over four publically available benchmark datasets, demonstrating the superiority of our work.

With the goal of effectively identifying common and salient objects in a group of relevant images, co-saliency detection has become essential for many applications such as video foreground extraction, surveillance, image retrieval, and image annotation. In this paper, we propose a unified co-saliency detection framework by introducing two novel insights: 1) looking deep to transfer higher-level representations by using the convolutional neural network with additional adaptive layers could better reflect the properties of the co-salient objects, especially their consistency among the image group; 2) looking wide to take advantage of the visually similar neighbors beyond a certain image group could effectively suppress the influence of the common background regions when formulating the intra-group consistency. In the proposed framework, the wide and deep information are explored for the object proposal windows extracted in each image, and the co-saliency scores are calculated by integrating the intra-image contrast and intra-group consistency via a principled Bayesian formulation. Finally the window-level co-saliency scores are converted to the superpixel-level co-saliency maps through a foreground region agreement strategy. Comprehensive experiments on two benchmark datasets have demonstrated the consistent performance gain of the proposed approach.

As an interesting and emerging topic, co-saliency detection aims at simultaneously extracting common salient objects in a group of images. Traditional co-saliency detection approaches rely heavily on human knowledge for designing hand-crafted metrics to explore the intrinsic patterns underlying co-salient objects. Such strategies, however, always suffer from poor generalization capability to flexibly adapt various scenarios in real applications, especially due to their lack of insightful understanding of the biological mechanisms of human visual co-attention. To alleviate this problem, we propose a novel framework for this task, by naturally reformulating it as a multiple-instance learning (MIL) problem and further integrating it into a self-paced learning (SPL) regime. The proposed framework on one hand is capable of fitting insightful metric measurements and discovering common patterns under co-salient regions in a self-learning way by MIL, and on the other hand tends to promise the learning reliability and stability by simulating the human learning process through SPL. Experiments on benchmark datasets have demonstrated the effectiveness of the proposed framework as compared with the state-of-the-arts.

In this paper, we focus on tacking the problem of weakly supervised semantic segmentation. The aim is to predict the class label of image regions under weakly supervised settings, where training images are only provided with image-level labels indicating the classes they contain. The main difficulty of weakly supervised semantic segmentation arises from the complex diversity of visual classes and the lack of supervision information for learning a multi-classes classifier. To conquer the challenge, we propose a novel discriminative deep feature learning framework based on stacked autoencoders (SAE) by integrating pairwise constraints to serve as a discriminative term. Furthermore, to mine effective supervision information, global context about co-occurrence of visual classes as well as local context around each image region is exploited as constraints for training a multi-class classifier. Finally, the classifier training is formulated as an ultimate optimization problem, which can be solved efficiently by an alternate iterative optimization method. Comprehensive experiments on the MSRC 21 dataset demonstrate the superior performance compared with several state-of-the-art weakly supervised image segmentation methods.

Effective representation and classification of scenes using very high resolution (VHR) remote sensing images cover a wide range of applications. Although robust low‐level image features have been proven to be effective for scene classification, they are not semantically meaningful and thus have difficulty to deal with challenging visual recognition tasks. In this study, the authors propose a new and effective auto‐encoder‐based method to learn a shared mid‐level visual dictionary. This dictionary serves as a shared and universal basis to discover mid‐level visual elements. On the one hand, the mid‐level visual dictionary learnt using machine learning technique is more discriminative and contains rich semantic information, compared with the traditional low‐level visual words. On the other hand, the mid‐level visual dictionary is more robust to occlusions and image clutters. In the authors' scene‐classification scheme, they use discriminative mid‐level visual elements, rather than individual pixels or low‐level image features, to represent images. This new image representation is able to capture much of the high‐level meaning and contents of the image, facilitating challenging remote sensing image scene‐classification tasks. Comprehensive evaluations on a challenging VHR remote sensing images data set and comparisons with state‐of‐the‐art approaches demonstrate the effectiveness and superiority of their study.

In this letter, we develop a novel framework of leveraging weakly supervised learning techniques to efficiently detect targets from remote sensing images, which enables us to reduce the tedious manual annotation for collecting training data while maintaining the detection accuracy to large extent. The proposed framework consists of a weakly supervised training procedure to yield the detectors and an effective scheme to detect targets from testing images. Comprehensive evaluations on three benchmarks which have different spatial resolutions and contain different types of targets as well as the comparisons with traditional supervised learning schemes demonstrate the efficiency and effectiveness of the proposed framework.

As the indicator of emotion intensity, arousal is a significant clue for users to find their interested content. Hence, effective techniques for video arousal recognition are highly required. In this paper, we propose a novel framework for recognizing arousal levels by integrating low-level audio-visual features derived from video content and human brain's functional activity in response to videos measured by functional magnetic resonance imaging (fMRI). At first, a set of audio-visual features which have been demonstrated to be correlated with video arousal are extracted. Then, the fMRI-derived features that convey the brain activity of comprehending videos are extracted based on a number of brain regions of interests (ROIs) identified by a universal brain reference system. Finally, these two sets of features are integrated to learn a joint representation by using a multimodal deep Boltzmann machine (DBM). The learned joint representation can be utilized as the feature for training classifiers. Due to the fact that fMRI scanning is expensive and time-consuming, our DBM fusion model has the ability to predict the joint representation of the videos without fMRI scans. The experimental results on a video benchmark demonstrated the effectiveness of our framework and the superiority of integrated features.

Detection of salient objects from images is gaining increasing research interest in recent years as it can substantially facilitate a wide range of content-based multimedia applications. Based on the assumption that foreground salient regions are distinctive within a certain context, most conventional approaches rely on a number of hand-designed features and their distinctiveness is measured using local or global contrast. Although these approaches have been shown to be effective in dealing with simple images, their limited capability may cause difficulties when dealing with more complicated images. This paper proposes a novel framework for saliency detection by first modeling the background and then separating salient objects from the background. We develop stacked denoising autoencoders with deep learning architectures to model the background where latent patterns are explored and more powerful representations of data are learned in an unsupervised and bottom-up manner. Afterward, we formulate the separation of salient objects from the background as a problem of measuring reconstruction residuals of deep autoencoders. Comprehensive evaluations of three benchmark datasets and comparisons with nine state-of-the-art algorithms demonstrate the superiority of this paper.

Generally, various visual media are unequally memorable by the human brain. This paper looks into a new direction of modeling the memorability of video clips and automatically predicting how memorable they are by learning from brain functional magnetic resonance imaging (fMRI). We propose a novel computational framework by integrating the power of low-level audiovisual features and brain activity decoding via fMRI. Initially, a user study experiment is performed to create a ground truth database for measuring video memorability and a set of effective low-level audiovisual features is examined in this database. Then, human subjects' brain fMRI data are obtained when they are watching the video clips. The fMRI-derived features that convey the brain activity of memorizing videos are extracted using a universal brain reference system. Finally, due to the fact that fMRI scanning is expensive and time-consuming, a computational model is learned on our benchmark dataset with the objective of maximizing the correlation between the low-level audiovisual features and the fMRI-derived features using joint subspace learning. The learned model can then automatically predict the memorability of videos without fMRI scans. Evaluations on publically available image and video databases demonstrate the effectiveness of the proposed framework.

Land-use classification using remote sensing images covers a wide range of applications. With more detailed spatial and textural information provided in very high resolution (VHR) remote sensing images, a greater range of objects and spatial patterns can be observed than ever before. This offers us a new opportunity for advancing the performance of land-use classification. In this paper, we first introduce an effective midlevel visual elementsoriented land-use classification method based on “partlets,” which are a library of pretrained part detectors used for midlevel visual elements discovery. Taking advantage of midlevel visual elements rather than low-level image features, a partlets-based method represents images by computing their responses to a large number of part detectors. As the number of part detectors grows, a main obstacle to the broader application of this method is its computational cost. To address this problem, we next propose a novel framework to train coarse-to-fine shared intermediate representations, which are termed “sparselets,” from a large number of pretrained part detectors. This is achieved by building a single-hidden-layer autoencoder and a single-hidden-layer neural network with an L0-norm sparsity constraint, respectively. Comprehensive evaluations on a publicly available 21-class VHR landuse data set and comparisons with state-of-the-art approaches demonstrate the effectiveness and superiority of this paper.

The abundant spatial and contextual information provided by the advanced remote sensing technology has facilitated subsequent automatic interpretation of the optical remote sensing images (RSIs). In this paper, a novel and effective geospatial object detection framework is proposed by combining the weakly supervised learning (WSL) and high-level feature learning. First, deep Boltzmann machine is adopted to infer the spatial and structural information encoded in the low-level and middle-level features to effectively describe objects in optical RSIs. Then, a novel WSL approach is presented to object detection where the training sets require only binary labels indicating whether an image contains the target object or not. Based on the learnt high-level features, it jointly integrates saliency, intraclass compactness, and interclass separability in a Bayesian framework to initialize a set of training examples from weakly labeled images and start iterative learning of the object detector. A novel evaluation criterion is also developed to detect model drift and cease the iterative learning. Comprehensive experiments on three optical RSI data sets have demonstrated the efficacy of the proposed approach in benchmarking with several state-of-the-art supervised-learning-based object detection approaches.

Feature extraction is of high importance for effective data classification in hyperspectral imaging (HSI). Considering the high correlation among band images, spectral-domain feature extraction is widely employed. For effective spatial information extraction, a 2-D extension to singular spectrum analysis (2D-SSA), which is a recent technique for generic data mining and temporal signal analysis, is proposed. With 2D-SSA applied to HSI, each band image is decomposed into varying trends, oscillations, and noise. Using the trend and the selected oscillations as features, the reconstructed signal, with noise highly suppressed, becomes more robust and effective for data classification. Three publicly available data sets for HSI remote sensing data classification are used in our experiments. Comprehensive results using a support vector machine classifier have quantitatively evaluated the efficacy of the proposed approach. Benchmarked with several state-of-the-art methods including 2-D empirical mode decomposition (2D-EMD), it is found that our proposed 2D-SSA approach generates the best results in most cases. Unlike 2D-EMD that requires sequential transforms to obtain detailed decomposition, 2D-SSA extracts all components simultaneously. As a result, the execution time in feature extraction can be also dramatically reduced. The superiority in terms of enhanced discrimination ability from 2D-SSA is further validated when a relatively weak classifier, i.e., the k-nearest neighbor, is used for data classification. In addition, the combination of 2D-SSA with 1-D principal component analysis (2D-SSA-PCA) has generated the best results among several other approaches, demonstrating the great potential in combining 2D-SSA with other approaches for effective spatial-spectral feature extraction and dimension reduction in HSI.

Nowadays, locations of images have been widely used in many application scenarios for large geo-tagged image corpora. As to images which are not geographically tagged, we estimate their locations with the help of the large geo-tagged image set by content-based image retrieval. In this paper, we exploit spatial information of useful visual words to improve image location estimation (or content-based image retrieval performances). We proposed to generate visual word groups by mean-shift clustering. To improve the retrieval performance, spatial constraint is utilized to code the relative position of visual words. We proposed to generate a position descriptor for each visual word and build fast indexing structure for visual word groups. Experiments show the effectiveness of our proposed approach.

Task-based fMRI (tfMRI) has been widely used to explore functional brain networks via predefined stimulus paradigm in the fMRI scan. Traditionally, the general linear model (GLM) has been a dominant approach to detect task-evoked networks. However, GLM focuses on task-evoked or event-evoked brain responses and possibly ignores the intrinsic brain functions. In comparison, dictionary learning and sparse coding methods have attracted much attention recently, and these methods have shown the promise of automatically and systematically decomposing fMRI signals into meaningful task-evoked and intrinsic concurrent networks. Nevertheless, two notable limitations of current data-driven dictionary learning method are that the prior knowledge of task paradigm is not sufficiently utilized and that the establishment of correspondences among dictionary atoms in different brains have been challenging. In this paper, we propose a novel supervised dictionary learning and sparse coding method for inferring functional networks from tfMRI data, which takes both of the advantages of model-driven method and data-driven method. The basic idea is to fix the task stimulus curves as predefined model-driven dictionary atoms and only optimize the other portion of data-driven dictionary atoms. Application of this novel methodology on the publicly available human connectome project (HCP) tfMRI datasets has achieved promising results.

In this paper, the salient region detection problem is investigated by using background contrast. Since background colors usually appear near image border, and all background colors can be mainly represented by the colors in the image boundary, the boundary-based model is established via computing the different between the intern colors and the boundary colors. As we know, the nearer the patches are close to center, the more they affect other patches. Based on this, a new distribution-based model is proposed. Because of the fact that pixels in the small neighborhood usually have the very similar color components, and computing region-based contrast can reduce the computational complexity, the superpixel algorithm is used in the pretreatment process. Finally, experimental results demonstrate that the proposed method outperforms the state-of-the-art approaches.

We here study the problem of visual attention computation in video of driving environment via the learning from eye movements. We collect a large-scale database of eye movements from 28 subjects on 30 videos of road scenes, which simulate the driving environment. The analysis on this eye movement database reveals that visual attention in driving environment is directed by high-level cognitive factors such as objects. We then present a new high-level representation called Traffic Object Bank (TOB), which is comprised of many individual road object detectors trained comprehensively in semantic space as well as viewpoint space. TOB provides semantically rich object-level features. Finally, we develop a computational model to predict where drivers look via the mapping from TOB-based representation and to gaze data. Experimental results on our traffic scene video benchmark indicate high accordance with human eye movement and show great promise for further applications.

Saliency detection has been a very active research area in recent years. Most traditional methods suffer from the problem that existing visual features are not discriminative or not robust enough to predict salient locations. As a result, the experimental results of these previous methods are still far from satisfactory. In this paper, we propose to utilize a two-layer Deep Boltzmann Machine (DBM) to learn enhanced features from existing contrast-based low-level features, which are more discriminative and reliable. A saliency computation model is then trained to build a mapping from those enhanced features to eye fixation data. The proposed work is amongst the earliest efforts of examining the feasibility of applying deep learning algorithms to saliency detection. Comprehensive evaluations on two publically available benchmark datasets and comparisons with a number of state-of-the-art approaches demonstrate the effectiveness of the proposed work.

In this paper we present a conceptually simple but surprisingly effective multi-class geospatial object detection method based on Collection of Part Detectors (COPD), which can be easily scaled to a larger number of object classes. The presented COPD is composed of a set of representative and discriminative part detectors, where each part detector is a linear support vector machine (SVM) classifier trained using a weakly supervised learning method that only requires image labels indicating the presence of objects for the training data. Here, each part detector corresponds to a particular viewpoint of an object class, so the collection of them provides a feasible solution for rotation-invariant and simultaneous detection of multi-class geospatial objects. Comprehensive evaluations on high-spatial-resolution remote sensing images and comparisons with a number of state-of-the-art approaches demonstrate the effectiveness and superiority of the presented method.

In this paper, we propose a 3D finger detecting and tracking algorithm (3dFDT) based on a sliding window for RGB-D sequences. Microsoft Kinect is utilized as a 3D depth camera, which provides RGB images and depth data simultaneously. However the depth data have large noises when the background is white. We are focusing on tracking the up or down movements of the fingertips in order to apply in some applications such as playing the piano on the table or " in the air". First, the skin statistical ellipse model is applied to detect the hand skin region, and the pseudo-contours are removed to get the interested hand region. Then, the region center is considered as the center of the palm. Next, the fingertips are located precisely using the convex defect detection based on the refined hand contours. Furthermore, the depth information is aligned to the fingertips in the RGB channels. Finally, the sliding window strategy is employed to stabilize the depth information due to large noises from the depth space especially when dealing with the white surface as backgrounds. The experimental results show that our proposed method is effective, and it can be applied in the real-time applications for non-contact interactions.

Given the growing number of available audio streams through a variety of sources and distribution channels, effective and advanced computational audio analysis has received increasing interest in the multimedia field. However, the effectiveness of current audio analysis strategies might be hampered due to the lack of effective representation of high-level semantics perceived by the human and the lack of effective approaches to bridging the gaps between most low-level acoustic features and high-level semantic features. This semantic gap has become the 'bottleneck' problem in audio analysis. In this paper, we propose a computational framework to decode biologically-plausible auditory saliency using high-level features derived from functional magnetic resonance imaging (fMRI) which monitors the human brain's response under the natural stimulus of audio listening. Specifically, we identify meaningful intrinsic brain networks which are involved in audio listening via effective online dictionary learning and sparse representation of whole-brain fMRI signals, reconstruct auditory saliency features using those identified brain network components, and perform group-wise analysis to identify consistent 'brain decoders' of the saliency features across different excerpts and participants. Experimental results demonstrate that the auditory saliency features are effectively decoded via our methods, which potentially provide opportunities for various applications in the multimedia field.

Abstract Since indoor scenes are frequently changed in daily life, such as re‐layout of furniture, the 3D reconstructions for them should be flexible and easy to update. We present an automatic 3D scene update algorithm to indoor scenes by capturing scene variation with RGBD cameras. We assume an initial scene has been reconstructed in advance in manual or other semi‐automatic way before the change, and automatically update the reconstruction according to the newly captured RGBD images of the real scene update. It starts with an automatic segmentation process without manual interaction, which benefits from accurate labeling training from the initial 3D scene. After the segmentation, objects captured by RGBD camera are extracted to form a local updated scene. We formulate an optimization problem to compare to the initial scene to locate moved objects. The moved objects are then integrated with static objects in the initial scene to generate a new 3D scene. We demonstrate the efficiency and robustness of our approach by updating the 3D scene of several real‐world scenes.

Content‐based image retrieval (CBIR) systems often incorporate a relevance feedback mechanism in which retrieval is adapted based on users identifying images as relevant or irrelevant. Such relevance decisions are often assumed to be category‐based. However, forcing a user to decide upon category membership of an image, even when unfamiliar with a database and irrespective of context, is restrictive. An alternative is to obtain user feedback in the form of relative similarity judgments. The ability of a user to provide meaningful feedback depends on the interface that displays retrieved images and facilitates the feedback. Similarity‐based 2D layouts provide context and can enable more efficient visual search. Motivated by these observations, this study describes and evaluates an interactive image browsing and retrieval approach based on relative similarity feedback obtained from 2D image layouts. It incorporates online maximal‐margin learning to adapt the image similarity metric used to perform retrieval. A user starts a session by browsing a collection of images displayed in a 2D layout. He/she may choose a query image perceived to be similar to the envisioned target image. A set of images similar to the query are then returned. The user can then provide relational feedback and/or update the query image to obtain a new set of images. Algorithms for CBIR are often characterised empirically by simulating usage based on pre‐defined, fixed category labels, deeming retrieved results as relevant if they share a category label with the query. In contrast, the purpose of the system in this study is to enable browsing and retrieval without predefined categories. Therefore evaluation is performed in a target‐based setting by quantifying the efficiency with which target images are retrieved given initial queries.

Neuroimaging and brain mapping can provide meaningful guidance to multimedia analyses. and advanced computational multimedia analysis can be used to better understand the functional mechanisms of the human brain. Essentially, brain imaging and brain mapping techniques can serve as a bridge that links the digital representation of multimedia and the perception and comprehension of its content. This paper summarizes methods that integrate brain imaging with multimedia analysis and discusses the opportunities and challenges in this interdisciplinary field. In general, quantitative modeling of brain responses during multimedia comprehension has advanced content-based multimedia studies such as image and video classification and tagging. Multimedia analysis has promoted functional brain mapping by using naturalistic multimedia as stimuli during neuroimaging. Challenges and opportunities in merging neuroimaging and multimedia include the quantification of the brain's responses, the quantification of multimedia, and the mapping between brain responses and computational multimedia features.

3-D shape analysis has attracted extensive research efforts in recent years, where the major challenge lies in designing an effective high-level 3-D shape feature. In this paper, we propose a multi-level 3-D shape feature extraction framework by using deep learning. The low-level 3-D shape descriptors are first encoded into geometric bag-of-words, from which middle-level patterns are discovered to explore geometric relationships among words. After that, high-level shape features are learned via deep belief networks, which are more discriminative for the tasks of shape classification and retrieval. Experiments on 3-D shape recognition and retrieval demonstrate the superior performance of the proposed method in comparison to the state-of-the-art methods.

This paper presents a novel structural model based scene recognition method. In order to resolve regular grid image division methods which cause low content discriminability for scene recognition in previous methods, we partition an image into a pre-defined set of regions by superpixel segmentation. And then classification is modelled by introducing a structural model which has the capability of organizing unordered features of image patches. In the implementation, CENTRIST which is robust to scene recognition is used as original image feature, and bag-of-words representation is used to capture the local appearances of an image. In addition, we incorporate adjacent superpixel's differences as edge features. Our models are trained using structural SVM. Two state-of-the-art scene datasets are adopted to evaluate the proposed method. The experiment results show that the recognition accuracy is significantly improved by the proposed method.

This study addresses the problem of vision‐based sign language recognition, which is to translate signs to English. The authors propose a fully automatic system that starts with breaking up signs into manageable subunits. A variety of spatiotemporal descriptors are extracted to form a feature vector for each subunit. Based on the obtained features, subunits are clustered to yield codebooks. A boosting algorithm is then applied to learn a subset of weak classifiers representing discriminative combinations of features and subunits, and to combine them into a strong classifier for each sign. A joint learning strategy is also adopted to share subunits across sign classes, which leads to a more efficient classification. Experimental results on real‐world hand gesture videos demonstrate the proposed approach is promising to build an effective and scalable system.

Saliency detection aims at quantitatively predicting attended locations in an image. It may mimic the selection mechanism of the human vision system, which processes a small subset of a massive amount of visual input while the redundant information is ignored. Motivated by the biological evidence that the receptive fields of simple cells in V1 of the vision system are similar to sparse codes learned from natural images, this paper proposes a novel framework for saliency detection by using image sparse coding representations as features. Unlike many previous approaches dedicated to examining the local or global contrast of each individual location, this paper develops a probabilistic computational algorithm by integrating objectness likelihood with appearance rarity. In the proposed framework, image sparse coding representations are yielded through learning on a large amount of eye-fixation patches from an eye-tracking dataset. The objectness likelihood is measured by three generic cues called compactness, continuity, and center bias. The appearance rarity is inferred by using a Gaussian mixture model. The proposed paper can serve as a basis for many techniques such as image/video segmentation, retrieval, retargeting, and compression. Extensive evaluations on benchmark databases and comparisons with a number of up-to-date algorithms demonstrate its effectiveness.

Meaningful representation and effective retrieval of video shots in a large-scale database has been a profound challenge for the image/video processing and computer vision communities. A great deal of effort has been devoted to the extraction of low-level visual features, such as color, shape, texture, and motion for characterizing and retrieving video shots. However, the accuracy of these feature descriptors is still far from satisfaction due to the well-known semantic gap. In order to alleviate the problem, this paper investigates a novel methodology of representing and retrieving video shots using human-centric high-level features derived in brain imaging space (BIS) where brain responses to natural stimulus of video watching can be explored and interpreted. At first, our recently developed dense individualized and common connectivity-based cortical landmarks (DICCCOL) system is employed to locate large-scale functional brain networks and their regions of interests (ROIs) that are involved in the comprehension of video stimulus. Then, functional connectivities between various functional ROI pairs are utilized as BIS features to characterize the brain's comprehension of video semantics. Then an effective feature selection procedure is applied to learn the most relevant features while removing redundancy, which results in the formation of the final BIS features. Afterwards, a mapping from low-level visual features to high-level semantic features in the BIS is built via the Gaussian process regression (GPR) algorithm, and a manifold structure is then inferred, in which video key frames are represented by the mapped feature vectors in the BIS. Finally, the manifold-ranking algorithm concerning the relationship among all data is applied to measure the similarity between key frames of video shots. Experimental results on the TRECVID 2005 dataset demonstrate the superiority of the proposed work in comparison with traditional methods.

With the availability of large amount of audio tracks through a variety of sources and distribution channels, automatic music/speech classification becomes an indispensable tool in social audio websites and online audio communities. However, the accuracy of current acoustic-based low-level feature classification methods is still rather far from satisfaction. The discrepancy between the limited descriptive power of low-level features and the richness of high-level semantics perceived by the human brain has become the 'bottleneck' problem in audio signal analysis. In this paper, functional magnetic resonance imaging (fMRI) which monitors the human brain's response under the natural stimulus of music/speech listening is used as high-level features in the brain imaging space (BIS). We developed a computational framework to model the relationships between BIS features and low-level features in the training dataset with fMRI scans, predict BIS features of testing dataset without fMRI scans, and use the predicted BIS features for music/speech classification in the application stage. Experimental results demonstrated the significantly improved performance of music/speech classification via predicted BIS features than that via the original low-level features.

The multimedia content analysis community has made significant efforts to bridge the gaps between low-level features and high-level semantics perceived by humans. Recent advances in brain imaging and neuroscience in exploring the human brain's responses during multimedia comprehension demonstrated the possibility of leveraging cognitive neuroscience knowledge to bridge the semantic gaps. This paper presents our initial effort in this direction by using functional magnetic resonance imaging (fMRI). Specifically, task-based fMRI (T-fMRI) was performed to accurately localize the brain regions involved in video comprehension. Then, natural stimulus fMRI (N-fMRI) data were acquired when subjects watched the multimedia clips selected from the TRECVID datasets. The responses in the localized brain regions were measured and used to extract high-level features as the representation of the brain's comprehension of semantics in the videos. A novel computational framework was developed to learn the most relevant low-level feature sets that best correlate the fMRI-derived semantic features based on the training videos with fMRI scans, and then the learned model was applied to larger scale TRECVID video datasets without fMRI scans for category classification. Our experimental results demonstrate: 1) there are meaningful couplings between brain's fMRI-derived responses and video stimuli, suggesting the validity of linking semantics and low-level features via fMRI and 2) the computationally learned low-level features can significantly (p <; 0.01) improve video classification in comparison with original low-level features and extracted low-level features resulted from well-known feature projection algorithms.

In recent two decades, a large amount of effort has been devoted to content-based video retrieval (CBVR), which aims to manage large-scale video databases in an effective way based on visual features such as color, shape, texture, and motion. However, the performance of CBVR systems is still far from satisfaction due to the well-known semantic gap. In order to alleviate the problem, this paper proposes a novel retrieval methodology using semantic features derived from brain imaging space (BIS) that reflects brain responses and interactions under natural stimulus of video watching. A mapping from visual features to semantic features in BIS is built through Gaussian process regression. A manifold structure is then inferred where video key frames are represented by mapped feature vectors in BIS. Finally, the manifold-ranking algorithm concerning the relationship among all data is applied to measure the similarity between key frames. Preliminary experimental results on the TRECVID 2005 dataset demonstrate the superiority of the proposed work in comparison with traditional methods.

Image saliency detection provides a powerful tool for predicting where human tends to look at in an image, which has been a long attempt for the computer vision community. In this paper, we propose a biologically-inspired model for computing image saliency. At first, a set of basis functions that accords with visual responses to natural stimuli is learned by using eye-fixation patches from an eye-tracking dataset. Three features are then derived based on the learned basis functions including continuity, clutter contrast, and local contrast. Finally, these three features are combined into the saliency map. The proposed approach is easy to implement and can be used in many image and video content analysis applications. Experiments on a large-scale benchmark dataset and comparisons with a number of the state-of-the-art approaches demonstrate its superiority.

The multimedia content analysis community has made significant effort to bridge the gap between low-level features and high-level semantics perceived by human cognitive systems such as real-world objects and concepts. In the two fields of multimedia analysis and brain imaging, both topics of low-level features and high level semantics are extensively studied. For instance, in the multimedia analysis field, many algorithms are available for multimedia feature extraction, and benchmark datasets are available such as the TRECVID. In the brain imaging field, brain regions that are responsible for vision, auditory perception, language, and working memory are well studied via functional magnetic resonance imaging (fMRI). This paper presents our initial effort in marrying these two fields in order to bridge the gaps between low-level features and high-level semantics via fMRI brain imaging. Our experimental paradigm is that we performed fMRI brain imaging when university student subjects watched the video clips selected from the TRECVID datasets. At current stage, we focus on the three concepts of sports, weather, and commercial-/advertisement specified in the TRECVID 2005. Meanwhile, the brain regions in vision, auditory, language, and working memory networks are quantitatively localized and mapped via task-based paradigm fMRI, and the fMRI responses in these regions are used to extract features as the representation of the brain's comprehension of semantics. Our computational framework aims to learn the most relevant low-level feature sets that best correlate the fMRI-derived semantics based on the training videos with fMRI scans, and then the learned models are applied to larger scale test datasets without fMRI scans for category classifications. Our result shows that: 1) there are meaningful couplings between brain's fMRI responses and video stimuli, suggesting the validity of linking semantics and low-level features via fMRI; 2) The computationally learned low-level feature sets from fMRI-derived semantic features can significantly improve the classification of video categories in comparison with that based on original low-level features.

Mechanisms for visualizing image collections are essential for browsing and exploring their content. This is especially true when metadata are ineffective in retrieving items due to the sparsity or esoteric nature of text. An obvious approach is to automatically lay out sets of images in ways that reflect relationships between the items. However, dimensionality reduction methods that map from high-dimensional content-based feature distributions to low-dimensional layout spaces for visualization often result in displays in which many items are occluded whilst large regions are empty or only sparsely populated. Furthermore, such methods do not consider the shape of the region of layout space to be populated. This paper proposes a method, high-entropy layout distributions. that addresses these limitations. Layout distributions with low differential entropy are penalized. An optimization strategy is presented that finds layouts that have high differential entropy and that reflect inter-image similarities. Efficient optimization is obtained using a step-size constraint and an approximation to quadratic (Renyi) entropy. Two image archives of cultural and commercial importance are used to illustrate and evaluate the method. A comparison with related methods demonstrates its effectiveness.

Recently, a promising research direction has emerged in sign language recognition (SLR) aimed at breaking up signs into manageable subunits. This paper presents a novel SL learning technique based on boosted subunits. Three main contributions distinguish the proposed work from traditional approaches: 1) A novel boosting framework is developed to recognize SL. The learning is based on subunits instead of the whole sign, which is more scalable for the recognition task. 2) Feature selection is performed to learn a small set of discriminative combinations of subunits and SL features. 3) A joint learning strategy is adopted to share subunits across sign classes, which leads to a better performance classifiers. Our experiments show that compared to Dynamic Time Warping (DTW) when applied on the whole sign, our proposed technique gives better results.

An approach to the analysis of images of regular texture is proposed in which lattice hypotheses are used to define statistical models. These models are then compared in terms of their ability to explain the image. A method based on this approach is described in which lattice hypotheses are generated using analysis of peaks in the image autocorrelation function, statistical models are based on Gaussian or Gaussian mixture clusters, and model comparison is performed using the marginal likelihood as approximated by the Bayes Information Criterion (BIC). Experiments on public domain regular texture images and a commercial textile image archive demonstrate substantially improved accuracy compared to two competing methods. The method is also used for classification of texture images as regular or irregular. An application to thumbnail image extraction is discussed. These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.

This paper proposes a generic model for unsupervised extraction of viewer's attention objects from color images. Without the full semantic understanding of image content, the model formulates the attention objects as a Markov random field (MRF) by integrating computational visual attention mechanisms with attention object growing techniques. Furthermore, we describe the MRF by a Gibbs random field with an energy function. The minimization of the energy function provides a practical way to obtain attention objects. Experimental results on 880 real images and user subjective evaluations by 16 subjects demonstrate the effectiveness of the proposed approach.

In MPEG-4, a visual scene may be treated as a composition of video objects and coded at object level. Such a flexible video coding framework makes it possible to code different video objects with different priority according to human perceptual characteristics. In this paper, we introduce a novel dynamic bit allocation framework to improve the subjective quality in such an object-based video coding system. We incorporate the rate distortion models with the dynamic priorities of the video objects and jointly encode video objects to minimize the weighted distortion within the bit budget constraint. We guarantee the human-interested video objects a better reconstructed quality by using the weighted bit allocation strategy in favour of the video objects with higher priority. To obtain the priority automatically, we apply a visual attention model. Comparing with traditional bit allocation algorithms, the objective quality of the object with higher priority is significantly improved under this framework. These results demonstrate the usefulness of this dynamic bit allocation framework.