Infrared small target detection is challenging due to limited target size and low signal-to-noise ratio. Unlike common targets, infrared small targets contain a higher proportion of edge pixels and exhibit blurred boundaries due to diffraction and quantization artifacts, making boundaries uniquely valuable cues for target perception. However, existing methods often emphasize holistic modeling while underutilizing such informative boundary cues. Motivated by this observation, we propose a Dual-Path Edge-Guided Frequency-Aware Network (DEFANet), which enables edge-target collaborative modeling for enhanced feature representation. DEFANet features a dual-path design, consisting of a main branch for holistic target modeling and an edge branch for boundary transition perception. To facilitate interaction and enhance representation in both branches, we introduce two core modules: Frequency-Aware Dual Enhancement Module (FADE) and Edge-Guided Integration Module (EGI). FADE employs a Frequency-Decoupled Attention Enhancement Mechanism to enhance both branches in the frequency domain, strengthening holistic modeling in the main branch and boundary representation in the edge branch. EGI leverages a Dual-Path Group-Wise Guidance Mechanism to integrate enhanced edge features into the main branch, improving boundary perception. Extensive experiments on four public infrared small target datasets, MDvsFA, LAFT, SIRST, and SIATD, demonstrate that DEFANet achieves SOTA performance. Ablation studies further validate the effectiveness of DEFANet and the soundness of its design motivation.

Multi-person eyeblink detection in untrimmed in-the-wild videos is a recently emerged and challenging task. Due to its significant spatio-temporal fine-grained characteristics compared to general actions, we empirically find that general action detectors, though effective in general domains, struggle with this task (i.e., Blink-AP < 2%). Specialized eyeblink detection methods alleviate it through fine-grained spatio-temporal operations. SOTA method proposes a unified model combining instance-aware face localization and eyeblink detection through joint multi-task learning and feature sharing. While effective, it exhibits two critical limitations that may contribute to its unsatisfactory performance (i.e., Blink-AP=10.11%): (1) Face localization and eyeblink detection require distinct spatio-temporal feature granularities, making joint modeling in a unified feature space suboptimal. (2) Eyeblink task training could be largely affected by unstable face-eye feature learning under the joint training paradigm. To address this, we propose DeFB, a decomposed feature learning paradigm with favorable effectiveness and efficiency: (1) We model faces and eyes in granularity-specific feature spaces, which enhances fine-grained perception while reducing computational costs compared to a unified feature space. (2) To mitigate face-eye feature learning instability, we adopt an asynchronous learning mechanism where eye feature learning refines well-trained coarse face features, with shared queries acting as a bridge between stages to retain the efficient feature sharing of existing unified models. Compared with SOTA method, DeFB doubles the performance (Blink-AP: 24.65% v.s. 10.11%) while boosting efficiency by nearly 35%. DeFB can also be integrated as a plug-in to substantially augment the eyeblink detection capabilities of general action detectors.

Bokeh rendering simulates the shallow depth-of-field effect in photography, enhancing visual aesthetics and guiding viewer attention to regions of interest. Although recent approaches perform well, rendering controllable bokeh without additional depth inputs remains a significant challenge. Existing classical and neural controllable methods rely on accurate depth maps, while generative approaches often struggle with limited controllability and efficiency. In this paper, we propose BokehFlow, a depth-free framework for controllable bokeh rendering based on flow matching. BokehFlow directly synthesizes photorealistic bokeh effects from all-in-focus images, eliminating the need for depth inputs. It employs a cross-attention mechanism to enable semantic control over both focus regions and blur intensity via text prompts. To support training and evaluation, we collect and synthesize four datasets. Extensive experiments demonstrate that BokehFlow achieves visually compelling bokeh effects and offers precise control, outperforming existing depth-dependent and generative methods in both rendering quality and efficiency.

Reconstructing high dynamic range (HDR) images from low dynamic range (LDR) bursts plays an essential role in the computational photography. Impressive progress has been achieved by learning-based algorithms which require LDR-HDR image pairs. However, these pairs are hard to obtain, which motivates researchers to delve into the problem of annotation-efficient HDR image reconstructing: how to achieve comparable performance with limited HDR ground truths (GTs). This work attempts to address this problem from the view of semi-supervised learning where a teacher model generates pseudo HDR GTs for the LDR samples without GTs and a student model learns from pseudo GTs. Nevertheless, the confirmation bias, i.e., the student may learn from the artifacts in pseudo HDR GTs, presents an impediment. To remove this impediment, an uncertainty-based masking process is proposed to discard unreliable parts of pseudo GTs at both pixel and patch levels, then the student can learn from the trusted areas. With this novel masking process, our semi-supervised HDR reconstructing method not only outperforms previous annotation-efficient algorithms, but also achieves comparable performance with up-to-date fully-supervised methods by using only 6.7% HDR GTs.

Bokeh is used in photography to emphasize the selected subject by smoothly blurring the out-of-focus region with appealing highlights. While recent advances have achieved impressive results in rendering realistic blur, existing frameworks typically rely on disparity maps and bokeh-relevant inputs (e.g., focal distance and blur size), and face significant challenges in video bokeh rendering due to limited temporal consistency. In this paper, we propose BokehCrafter, the first video diffusion framework that generates temporally coherent and visually pleasing bokeh effects from all-in-focus video inputs under user-friendly input conditions. Specifically, we leverage a dual-stream attention mechanism, integrating a reference image branch and a rendering instruction branch. We propose a Bokeh Image Extraction (BIE) module and a CLIP-based text encoder to extract image and text features, respectively, whose outputs are fused via a Text-Image Fusion (TIF) module to enable fine-grained and controllable bokeh rendering. To support the novel capabilities of our model, we construct Video Bokeh Scenes (VBS), a large-scale dataset containing a wide variety of bokeh videos with corresponding rendering instructions, across various scenes and rendering settings. Extensive experiments demonstrate that our method significantly outperforms state-of-the-art methods in both bokeh rendering quality and temporal consistency.

In this paper, we propose to address monocular 3D hand pose estimation from a single RGB or depth image via articulated anchor-to-joint 3D local regressors, in form of A2J-Transformer+. The key idea is to make the local regressors (i.e., anchor points) in 3D space be aware of hand's local fine details and global articulated context jointly, to facilitate predicting their 3D offsets toward hand joints with linear weighted aggregation for joint localization. Our intuition is that, local fine details help to estimate accurate offset but may suffer from the issues including serious occlusion, confusing similar patterns, and overfitting risk. On the other hand, hand's global articulated context can essentially provide additional descriptive clues and constraints to alleviate these issues. To set anchor points adaptively in 3D space, A2J-Transformer+ runs in a 2-stage manner. At the first stage, since the input modality property anchor points distribute more densely on X-Y plane, it leads to lower prediction accuracy along Z direction compared with those in the X and Y directions. To alleviate this, at the second stage anchor points are set near the joints yielded by the first stage evenly along X, Y, and Z directions. This treatment brings two main advantages: (1) balancing the prediction accuracy along X, Y, and Z directions, and (2) ensuring the anchor-joint offsets are of small values relatively easy to estimate. Wide-range experiments on three RGB hand datasets (InterHand2.6 M, HO-3D V2 and RHP) and three depth hand datasets (NYU, ICVL and HANDS 2017) verify A2J-Transformer+'s superiority and generalization ability for different modalities (i.e., RGB and depth) and hand cases (i.e., single hand, interacting hands, and hand-object interaction), even outperforming model-based manners. The test on ITOP dataset reveals that, A2J-Transformer+ can also be applied to 3D human pose estimation task.

Cross-view geo-localization (CVGL) determines the geographic location of a query image by matching it with the most similar GPS-tagged satellite images. Existing methods have improved cross-view feature consistency through complex network architectures but still face challenges under significant illumination variations. Inspired by cognitive psychology, we observe that humans can selectively ignore color distractions and focus on structural information under certain circumstances. Therefore, we propose an SPNet, leveraging sketches to enhance structural feature consistency and achieve illumination-robust CVGL. SPNet employs Sketch Proxy Training, which introduces both RGB and sketch images during training. By leveraging the structural consistency provided by sketches under color variations, it mitigates the network’s over-reliance on color cues and achieves robustness to illumination changes. To further integrate multi-source information, we design a Recurrent Progressive Channel Attention (RPCA) module that progressively selects and reweights channel features, effectively combining the semantic cues of RGB images with the structural information of sketches. In addition, we introduce a High-Pass Cross-Layer Connection (HPCL) to transmit high-frequency information across feature layers, emphasizing edges and details to reinforce structural modeling and cross-layer feature consistency. Extensive experiments demonstrate that our SPNet achieves superior performance on several CVGL datasets, including University-1652, SUES-200, CVUSA, and CVACT, and ranks first on the public leaderboard of the University-160k-WX dataset released by UAVM 2024, further validating its robustness and generalization capability.

The labeling difficulty has been a longstanding problem in deep image matting. To escape from fine labels, this work explores using rough annotations such as trimaps coarsely indicating the foreground/background as supervision. We present that the cooperation between learned semantics from indicated known regions and proper assumed matting rules can help infer alpha values at transition areas. Inspired by the nonlocal principle in traditional image matting, we build a directional distance consistency loss (DDC loss) at each pixel neighborhood to constrain the alpha values conditioned on the input image. DDC loss forces the distance of similar pairs on the alpha matte and on its corresponding image to be consistent. In this way, the alpha values can be propagated from learned known regions to unknown transition areas. With only images and trimaps, a matting model can be trained under the supervision of a known loss and the proposed DDC loss. Experiments on AM-2K and P3M-10K dataset show that our paradigm achieves comparable performance with the fine-label-supervised baseline, while sometimes offers even more satisfying results than human-labeled ground truth.

Existing research on avatar creation is typically limited to laboratory datasets, which require high costs against scalability and exhibit insufficient representation of the real world. On the other hand, the web abounds with off-the-shelf real- world human videos, but these videos vary in quality and require accurate annotations for avatar creation. To this end, we propose an automatic annotating pipeline with filtering protocols to curate these humans from the web. Our pipeline surpasses state-of-the-art methods on the EMDB benchmark, and the filtering protocols boost verification metrics on web videos. We then curate WildAvatar, a web-scale in-the-wild human avatar creation dataset extracted from YouTube, with 10,000+ different human subjects and scenes. WildAvatar is at least 10 × richer than previous datasets for 3D human avatar creation and closer to the real world. To explore its potential, we demonstrate the quality and generalizability of avatar creation methods on WildAvatar. We will publicly release our code, data source links and annotations to push forward 3D human avatar creation and other related fields for real-world applications.

Radar-Camera depth estimation aims to predict dense and accurate metric depth by fusing input images and Radar data. Model efficiency is crucial for this task in pursuit of real-time processing on autonomous vehicles and robotic platforms. However, due to the sparsity of Radar returns, the prevailing methods adopt multi-stage frameworks with intermediate quasi-dense depth, which are time-consuming and not robust. To address these challenges, we propose TacoDepth, an efficient and accurate Radar-Camera depth estimation model with one-stage fusion. Specifically, the graph-based Radar structure extractor and the pyramid-based Radar fusion module are designed to capture and integrate the graph structures of Radar point clouds, delivering superior model efficiency and robustness without relying on the intermediate depth results. Moreover, TacoDepth can be flexible for different inference modes, providing a better balance of speed and accuracy. Extensive experiments are conducted to demonstrate the efficacy of our method. Compared with the previous state-of-the-art approach, TacoDepth improves depth accuracy and processing speed by 12.8% and 91.8%. Our work provides a new perspective on efficient Radar-Camera depth estimation.

Depth estimation is a fundamental task in 3D vision. An ideal depth estimation model is expected to embrace meticulous detail, temporal consistency, and high efficiency. Although existing foundation models can perform well in certain specific aspects, most of them fall short of fulfilling all the above requirements simultaneously. In this paper, we present CH3Depth, an efficient and flexible model for depth estimation with flow matching to address this challenge. Specifically, 1) we reframe the optimization objective of flow matching as the Inversion by Direct Iteration (InDI) to improve accuracy. 2) To enhance efficiency, we propose non-uniform sampling to achieve better prediction with fewer sampling steps. 3) We design the Latent Temporal Stabilizer (LTS) to enhance temporal consistency by aggregating latent codes of adjacent frames, enabling our method to be lightweight and compatible for video depth estimation. CH3Depth achieves state-of-the-art performance in zero-shot evaluations across multiple image and video datasets, excelling in prediction accuracy, efficiency, and temporal consistency, highlighting its potential as the next foundation model for depth estimation.

Recent advances in 3D Gaussian Splatting (3D-GS) have shown remarkable success in representing 3D scenes and generating high-quality, novel views in real-time. However, 3D-GS and its variants assume that input images are captured based on pinhole imaging and are fully in focus. This assumption limits their applicability, as real-world images often feature shallow depth-of-field (DoF). In this paper, we introduce DoF-Gaussian, a controllable depth-of-field method for 3D-GS. We develop a lens-based imaging model based on geometric optics principles to control DoF effects. To ensure accurate scene geometry, we incorporate depth priors adjusted per scene, and we apply defocus-to-focus adaptation to minimize the gap in the circle of confusion. We also introduce a synthetic dataset to assess refocusing capabilities and the model’s ability to learn precise lens parameters. Our framework is customizable and supports various interactive applications. Extensive experiments confirm the effectiveness of our method. Our project is available at https://dof-gaussian.github.io/.

Referring expression counting (REC) algorithms are for more flexible and interactive counting ability across varied fine-grained text expressions. However, the requirement for fine-grained attribute understanding poses challenges for prior arts, as they struggle to accurately align attribute information with correct visual patterns. Given the proven importance of “visual density”, it is presumed that the limitations of current REC approaches stem from an under-exploration of “contextual attribute density” (CAD). In the scope of REC, we define CAD as the measure of the information intensity of one certain fine-grained attribute in visual regions. To model the CAD, we propose a U- shape CAD estimator in which referring expression and multi-scale visual features from GroundingDINO can interact with each other With additional density supervision, we can effectively encode CAD, which is subsequently decoded via a novel attention procedure with CAD-refined queries. Integrating all these contributions, our framework significantly outperforms state-of-the-art REC methods, achieves 30% error reduction in counting metrics and a 10% improvement in localization accuracy. The surprising results shed light on the significance of contextual attribute density for REC. Code will be at github.com/Xu3XiWang/CAD-GD.

We present Free4D, a novel tuning-free framework for 4D scene generation from a single image. Existing methods either focus on object-level generation, making scene-level generation infeasible, or rely on large-scale multi-view video datasets for expensive training, with limited generalization ability due to the scarcity of 4D scene data. In contrast, our key insight is to distill pre-trained foundation models for consistent 4D scene representation, which offers promising advantages such as efficiency and generalizability. 1) To achieve this, we first animate the input image using image-to-video diffusion models followed by 4D geometric structure initialization. 2) To turn this coarse structure into spatial-temporal consistent multiview videos, we design an adaptive guidance mechanism with a point-guided denoising strategy for spatial consistency and a novel latent replacement strategy for temporal coherence. 3) To lift these generated observations into consistent 4D representation, we propose a modulation-based refinement to mitigate inconsistencies while fully leveraging the generated information. The resulting 4D representation enables real-time, controllable rendering, marking a significant advancement in single-image-based 4D scene generation.

Video reframing, which converts landscape-oriented (LO) to portrait-oriented (PO) video for some PO devices such as smartphones and tablets, faces challenges. Existing approaches mainly follow a multi-step pipeline to preserve video content that ignore composition quality due to lack of large-scale datasets. To address these challenges, we propose a fully automated composition-aware dataset using vision-language models and image composition assessment models, pairing LO videos with high-quality PO versions. We then propose an end-to-end model with an attention-aware backbone and a time-aware consistency module. Experiments show our approach outperforms others in efficiency and effectiveness, proving that composition awareness and end-to-end modeling are critical for video reframing.

3D human pose lifting from a single RGB image is a challenging task in 3D vision. Existing methods typically establish a direct joint-to-joint mapping from 2D to 3D poses based on 2D features. This formulation suffers from two fundamental limitations: inevitable error propagation from input predicted 2D pose to 3D predictions and inherent difficulties in handling self-occlusion cases. In this paper, we propose PandaPose, a 3D human pose lifting approach via propagating 2D pose prior to 3D anchor space as the unified intermediate representation. Specifically, our 3D anchor space comprises: (1) Joint-wise 3D anchors in the canonical coordinate system, providing accurate and robust priors to mitigate 2D pose estimation inaccuracies. (2) Depth-aware joint-wise feature lifting that hierarchically integrates depth information to resolve self-occlusion ambiguities. (3) The anchor-feature interaction decoder that incorporates 3D anchors with lifted features to generate unified anchor queries encapsulating joint-wise 3D anchor set, visual cues and geometric depth information. The anchor queries are further employed to facilitate anchor-to-joint ensemble prediction. Experiments on three well-established benchmarks (i.e., Human3.6M, MPI-INF-3DHP and 3DPW) demonstrate the superiority of our proposition. The substantial reduction in error by 14.7% compared to SOTA methods on the challenging conditions of Human3.6M and qualitative comparisons further showcase the effectiveness and robustness of our approach.

We present the first framework capable of synthesizing the all-in-focus neural radiance field (NeRF) from inputs without manual refocusing. Without refocusing, the camera will automatically focus on the fixed object for all views, and current NeRF methods typically using one camera fail due to the consistent defocus blur and a lack of sharp reference. To restore the all-in-focus NeRF, we introduce the dual-camera from smartphones, where the ultra-wide camera has a wider depth-of-field (DoF) and the main camera possesses a higher resolution. The dual camera pair saves the high-fidelity details from the main camera and uses the ultra-wide camera's deep DoF as reference for all-in-focus restoration. To this end, we first implement spatial warping and color matching to align the dual camera, followed by a defocus-aware fusion module with learnable defocus parameters to predict a defocus map and fuse the aligned camera pair. We also build a multi-view dataset that includes image pairs of the main and ultra-wide cameras in a smartphone. Extensive experiments on this dataset verify that our solution, termed DC-NeRF, can produce high-quality all-in-focus novel views and compares favorably against strong baselines quantitatively and qualitatively. We further show DoF applications of DC-NeRF with adjustable blur intensity and focal plane, including refocusing and split diopter.

Despite significant advancements in simulating the bokeh effect of Digital Single Lens Reflex Camera (DSLR) from an all-in-focus image, challenges remain in processing highlight points, preserving boundary details for in-focus objects and processing high-resolution images efficiently. To tackle these issues, we first develop a ray-tracing-based bokeh simulator. An innovative pipeline with weight redistribution is introduced to handle highlight rendering. By considering the front length of lens barrel, we can simulate realistic cat-eye effect. This bokeh simulator serves as the foundation for creating our training dataset. Building on this dataset, we introduce a hybrid framework BokehMe++, combining a classical renderer and a neural renderer. The classical renderer is implemented by a hierarchical scattering-based method, which suffers from boundary inaccuracies. These erroneous areas will be identified by an error map generator and be corrected by a two-stage neural renderer. Adaptive resizing and iterative upsampling are introduced in the neural renderer to process arbitrary blur size efficiently. Extensive experiments demonstrate that BokehMe++ outperforms existing methods and provides highly customizable rendering features, such as adjustable blur amount, focal plane, highlight mode and cat-eye effect. Furthermore, BokehMe++ can maintain the sharpness of hair details in portraits through an auxiliary alpha map input.

Video depth estimation aims to infer temporally consistent depth. One approach is to finetune a single-image model on each video with geometry constraints, which proves inefficient and lacks robustness. An alternative is learning to enforce consistency from data, which requires well-designed models and sufficient video depth data. To address both challenges, we introduce NVDS that stabilizes inconsistent depth estimated by various single-image models in a plug-and-play manner. We also elaborate a large-scale Video Depth in the Wild (VDW) dataset, which contains 14,203 videos with over two million frames, making it the largest natural-scene video depth dataset. Additionally, a bidirectional inference strategy is designed to improve consistency by adaptively fusing forward and backward predictions. We instantiate a model family ranging from small to large scales for different applications. The method is evaluated on VDW dataset and three public benchmarks. To further prove the versatility, we extend NVDS to video semantic segmentation and several downstream applications like bokeh rendering, novel view synthesis, and 3D reconstruction. Experimental results show that our method achieves significant improvements in consistency, accuracy, and efficiency. Our work serves as a solid baseline and data foundation for learning-based video depth estimation.

Infrared small-target detection (ISTD) presents significant challenges due to the blurred edges and low signal-to-noise ratio (SNR) of the targets, which often results in the loss of critical target features during convolution and downsampling. While researchers have explored complex architectures to preserve these features, such designs often come at the cost of increased computational cost. Our analysis reveals that the curse of dimensionality (COD) is a fundamental issue in ISTD. Due to the sparse information content of infrared small targets, they struggle to effectively utilize high-dimensional feature spaces, leading to a significant feature loss phenomenon. To address this, we propose the CODNet, which integrates spatial–channel shuffle (SCS) attention, dynamic gated channel attention (DGCA), and DCT-based frequency attention (DFA) in the UNet framework. SCS enhances feature representation through spatial attention and channel shuffling, making the feature space more compact and informative. DGCA performs selective channel compression to improve the feature SNR while reducing redundancy and sparsity. DFA incorporates frequency-domain features to supplement spatial representations, alleviating feature sparsity and improving information utilization. Experimental results demonstrate that the CODNet achieves state-of-the-art performance on four public datasets while maintaining relatively high operation efficiency.

This work pays the first research effort to leverage point cloud sequence-based Self-supervised 3-D Action Feature Learning (S3AFL), under text's cross-modality weak supervision. We intend to fill the huge performance gap between point cloud sequence and 3-D skeleton-based manners. The key intuition derives from the observation that skeleton-based manners actually hold the human pose's high-level knowledge that leads to attention on the body's joint-aware local parts. Inspired by this, we propose to introduce the text's weak supervision of high-level semantics into a point cloud sequence-based paradigm. With RGB-point cloud pair sequence acquired via RGB-D camera, text sequence is first generated from RGB component using pretrained image captioning model, as auxiliary weak supervision. Then, S3AFL runs in a cross and intra-modality contrastive learning (CL) way. To resist text's missing and redundant semantics, feature learning is conducted in a multistage way with semantic refinement. Essentially, text is only required for training. To facilitate the feature's representation power on fine-grained actions, a multirank max-pooling (MR-MP) way is also proposed for the point set network to better maintain discriminative clues. Experiments verify that the text's weak supervision can facilitate performance by 10.8%, 10.4%, and 8.0% on NTU RGB+D 60, 120, and N-UCLA at most. The performance gap between point cloud sequence and skeleton-based manners has been remarkably narrowed down. The idea of transferring text's weak supervision to S3AFL can also be applied to a skeleton manner, with strong generality. The source code is available at https://github.com/tangent-T/W3AMT.

Novel view synthesis for dynamic 3D scenes poses a significant challenge. Many notable efforts use NeRF-based approaches to address this task and yield impressive results. However, these methods rely heavily on sufficient motion parallax in the input images or videos. When the camera motion range becomes limited or even stationary (i.e., small camera motion), existing methods encounter two primary challenges: incorrect representation of scene geometry and inaccurate estimation of camera parameters. These challenges make prior methods struggle to produce satisfactory results or even become ineffective. To address the first challenge, we propose a novel Distribution-based Depth Regularization (DDR) that ensures the rendering weight distribution to align with the true distribution. Specifically, unlike previous methods that use depth loss to calculate the error of the expectation, we calculate the expectation of the error by using Gumbel-softmax to differentiably sample points from discrete rendering weight distribution. Additionally, we introduce constraints that enforce the volume density of spatial points before the object boundary along the ray to be near zero, ensuring that our model learns the correct geometry of the scene. To demystify the DDR, we further propose a visualization tool that enables observing the scene geometry representation at the rendering weight level. For the second challenge, we incorporate camera parameter learning during training to enhance the robustness of our model to camera parameters. We conduct extensive experiments to demonstrate the effectiveness of our approach in representing scenes with small camera motion input, and our results compare favorably to state-of-the-art methods.

Class-agnostic motion prediction methods aim to comprehend motion within open-world scenarios, holding significance for autonomous driving systems. However, training a high-performance model in a fully-supervised manner always requires substantial amounts of manually annotated data, which can be both expensive and time-consuming to obtain. To address this challenge, our study explores the potential of semi-supervised learning (SSL) for class-agnostic motion prediction. Our SSL framework adopts a consistency-based self-training paradigm, enabling the model to learn from unlabeled data by generating pseudo labels through test-time inference. To improve the quality of pseudo labels, we propose a novel motion selection and re-generation module. This module effectively selects reliable pseudo labels and re-generates unreliable ones. Furthermore, we propose two data augmentation strategies: temporal sampling and BEVMix. These strategies facilitate consistency regularization in SSL. Experiments conducted on nuScenes demonstrate that our SSL method can surpass the self-supervised approach by a large margin by utilizing only a tiny fraction of labeled data. Furthermore, our method exhibits comparable performance to weakly and some fully supervised methods. These results highlight the ability of our method to strike a favorable balance between annotation costs and performance. Code will be available at https://github.com/kwwcv/SSMP.

Class-agnostic counting (CAC) aims to count objects of interest from a query image given few exemplars. This task is typically addressed by extracting the features of query image and exemplars respectively and then matching their feature similarity, leading to an extract-then-match paradigm. In this work, we show that CAC can be simplified in an extract-and-match manner, particularly using a vision transformer (ViT) where feature extraction and similarity matching are executed simultaneously within the self-attention. We reveal the rationale of such simplification from a decoupled view of the self-attention.The resulting model, termed CACViT, simplifies the CAC pipeline into a single pretrained plain ViT. Further, to compensate the loss of the scale and the order-of-magnitude information due to resizing and normalization in plain ViT, we present two effective strategies for scale and magnitude embedding. Extensive experiments on the FSC147 and the CARPK datasets show that CACViT significantly outperforms state-of-the-art CAC approaches in both effectiveness (23.60% error reduction) and generalization, which suggests CACViT provides a concise and strong baseline for CAC. Code will be available.

Generalizable NeRF aims to synthesize novel views for unseen scenes. Common practices involve constructing variance-based cost volumes for geometry reconstruction and encoding 3D descriptors for decoding novel views. However, existing methods show limited generalization ability in challenging conditions due to inaccurate geometry, sub-optimal descriptors, and decoding strategies. We address these issues point by point. First, we find the variance-based cost volume exhibits failure patterns as the features of pixels corresponding to the same point can be inconsistent across different views due to occlusions or reflections. We introduce an Adaptive Cost Aggregation (ACA) approach to amplify the contribution of consistent pixel pairs and suppress inconsistent ones. Un-like previous methods that solely fuse 2D features into descriptors, our approach introduces a Spatial-View Ag-gregator (SVA) to incorporate 3D context into descriptors through spatial and inter-view interaction. When decoding the descriptors, we observe the two existing decoding strategies excel in different areas, which are complementary. A Consistency-Aware Fusion (CAF) strategy is proposed to leverage the advantages of both. We incorporate the above ACA, SVA, and CAF into a coarse-to-fine framework, termed Geometry-aware Reconstruction and Fusion-refined Rendering (GeFu). GeFu attains state-of-the-art performance across multiple datasets. Code is available at https://github.com/TQTQliu/GeFu.

Current 3D stylization methods often assume static scenes, which violates the dynamic nature of our real world. To address this limitation, we present S-DyRF, a reference-based spatio-temporal stylization method for dynamic neu-ral radiance fields. However, stylizing dynamic 3D scenes is inherently challenging due to the limited availability of stylized reference images along the temporal axis. Our key insight lies in introducing additional temporal cues besides the provided reference. To this end, we generate temporal pseudo-references from the given stylized reference. These pseudo-references facilitate the propagation of style infor-mation from the reference to the entire dynamic 3D scene. For coarse style transfer, we enforce novel views and times to mimic the style details present in pseudo-references at the feature level. To preserve high-frequency details, we create a collection of stylized temporal pseudo-rays from temporal pseudo-references. These pseudo-rays serve as detailed and explicit stylization guidance for achieving fine style trans-fer. Experiments on both synthetic and real-world datasets demonstrate that our method yields plausible stylized re-sults of space-time view synthesis on dynamic 3D scenes.

In this paper, we present RStab, a novel framework for video stabilization that integrates 3D multi-frame fusion through volume rendering. Departing from conventional methods, we introduce a 3D multi-frame perspective to generate stabilized images, addressing the challenge of full-frame generation while preserving structure. The core of our RStab framework lies in Stabilized Rendering (SR), a volume rendering module, fusing multi-frame information in 3D space. Specifically, SR involves warping features and colors from multiple frames by projection, fusing them into descriptors to render the stabilized image. However, the precision of warped information depends on the projection accuracy, a factor significantly influenced by dynamic regions. In response, we introduce the Adaptive Ray Range (ARR) module to integrate depth priors, adaptively defining the sampling range for the projection process. Additionally, we propose Color Correction (CC) assisting geometric constraints with optical flow for accurate color aggregation. Thanks to the three modules, our RStab demonstrates superior performance compared with previous stabilizers in the field of view (FOV), image quality, and video stability across various datasets.

Recent advancements in dynamic neural radiance field methods have yielded remarkable outcomes. However, these approaches rely on the assumption of sharp input images. When faced with motion blur, existing dynamic NeRF methods often struggle to generate high-quality novel views. In this paper, we propose DyBluRF, a dynamic radiance field approach that synthesizes sharp novel views from a monocular video affected by motion blur. To account for motion blur in input images, we simultaneously capture the camera trajectory and object Discrete Cosine Transform (DCT) trajectories within the scene. Additionally, we employ a global cross-time rendering approach to ensure consistent temporal coherence across the entire scene. We curate a dataset comprising diverse dynamic scenes that are specifically tailored for our task. Experimental results on our dataset demonstrate that our method outperforms existing approaches in generating sharp novel views from motion-blurred inputs while maintaining spatial-temporal consistency of the scene.

The perception of motion behavior in a dynamic environment holds significant importance for autonomous driving systems, wherein class-agnostic motion prediction methods directly predict the motion of the entire point cloud. While most existing methods rely on fully-supervised learning, the manual labeling of point cloud data is laborious and time-consuming. Therefore, several annotation-efficient methods have been proposed to address this challenge. Al-though effective, these methods rely on weak annotations or additional multi-modal data like images, and the potential benefits inherent in the point cloud sequence are still underexplored. To this end, we explore the feasibility of self-supervised motion prediction with only unlabeled Li-DAR point clouds. Initially, we employ an optimal transport solver to establish coarse correspondences between current and future point clouds as the coarse pseudo motion labels. Training models directly using such coarse labels leads to noticeable spatial and temporal prediction in-consistencies. To mitigate these issues, we introduce three simple spatial and temporal regularization losses, which fa-cilitate the self-supervised training process effectively. Experimental results demonstrate the significant superiority of our approach over the state-of-the-art self-supervised methods. Code will be available at https://github.com/kwwcv/SelfMotion.

Automatic and interactive matting largely improve image matting by respectively alleviating the need for auxil-iary input and enabling object selection. Due to different settings on whether prompts exist, they either suffer from weakness in instance completeness or region details. Also, when dealing with different scenarios, directly switching between the two matting models introduces inconvenience and higher workload. Therefore, we wonder whether we can al-leviate the limitations of both settings while achieving unification to facilitate more convenient use. Our key idea is to offer saliency guidance for automatic mode to enable its attention to detailed regions, and also refine the instance completeness in interactive mode by replacing the binary mask guidance with a more probabilistic form. With different guidance for each mode, we can achieve unification through adaptable guidance, defined as saliency information in automatic mode and user cue for interactive one. It is instantiated as candidate feature in our method, an automatic switch for class token in pretrained ViTs and average feature of user prompts, controlled by the existence of user prompts. Then we use the candidate feature to generate a probabilistic similarity map as the guidance to alleviate the over-reliance on binary mask. Extensive experiments show that our method can adapt well to both automatic and inter-active scenarios with more light-weight framework. Code available at github.com/coconut/SMat.

We present the All-Seeing (AS) project: a large-scale dataset and model for recognizing and understanding everything in the open world. Using a scalable data engine that incorporates human feedback and efficient models in the loop, we create a new dataset (AS-1B) with over 1.2 billion regions annotated with semantic tags, question-answering pairs, and detailed captions. It covers a wide range of 3.5 million common and rare concepts in the real world and has 132.2 billion tokens that describe the concepts and their attributes. Leveraging this new dataset, we develop the All-Seeing model (ASM), a unified framework for panoptic visual recognition and understanding. The model is trained with open-ended language prompts and locations, which allows it to generalize to various vision and language tasks with remarkable zero-shot performance, including both region- and image-level retrieval, region recognition, captioning, and question-answering. We hope that this project can serve as a foundation for vision-language artificial general intelligence research. Code is available at https://github.com/OpenGVLab/all-seeing.

High dynamic range (HDR) video rendering from low dynamic range (LDR) videos where frames are of alternate exposures encounters significant challenges, due to the exposure change and absence at each time stamp. The exposure change and absence make existing methods generate flickering HDR results. In this paper, we propose a novel paradigm to render HDR frames via completing the absent exposure information, hence the exposure information is complete and consistent. Our approach involves interpolating neighbor LDR frames in the time dimension to reconstruct LDR frames for the absent exposures. Combining the interpolated and given LDR frames, the complete set of exposure information is available at each time stamp. This benefits the fusing process for HDR results, reducing noise and ghosting artifacts therefore improving temporal consistency. Extensive experimental evaluations on standard benchmarks demonstrate that our method achieves state-of-the-art performance, highlighting the importance of absent exposure completing in HDR video rendering. The code is available at https://github.com/cuijiahao666/NECHDR.

3D content creation has long been a complex and time-consuming process, often requiring specialized skills and resources. While re- cent advancements have allowed for text-guided 3D object and scene generation, they still fall short of providing sufficient control over the generation process, leading to a gap between the user’s creative vision and the generated results. In this paper, we present iControl3D, a novel interactive system that empowers users to gen- erate and render customizable 3D scenes with precise control. To this end, a 3D creator interface has been developed to provide users with fine-grained control over the creation process. Technically, we leverage 3D meshes as an intermediary proxy to iteratively merge individual 2D diffusion-generated images into a cohesive and uni- fied 3D scene representation. To ensure seamless integration of 3D meshes, we propose to perform boundary-aware depth alignment before fusing the newly generated mesh with the existing one in 3D space. Additionally, to effectively manage depth discrepancies between remote content and foreground, we propose to model re- mote content separately with an environment map instead of 3D meshes. Finally, our neural rendering interface enables users to build a radiance field of their scene online and navigate the entire scene. Extensive experiments have been conducted to demonstrate the effectiveness of our system. The code will be made available at https://github.com/xingyi- li/iControl3D.

Bokeh is a wide-aperture optical effect that creates aesthetic blurring in photography. However, achieving this effect typically demands expensive professional equipment and expertise. To make such cinematic techniques more accessible, bokeh rendering aims to generate the desired bokeh effects from all-in-focus inputs captured by smartphones. Previous efforts in bokeh rendering primarily focus on static images. However, when extended to video inputs, these methods exhibit flicker and artifacts due to a lack of temporal consistency modeling. Meanwhile, they cannot utilize information like occluded objects from adjacent frames, which are necessary for bokeh rendering. Moreover, the difficulties of capturing all-in-focus and bokeh video pairs result in a shortage of data for training video bokeh models. To tackle these challenges, we propose the Video Bokeh Renderer (VBR), the model designed specifically for video bokeh rendering.VBR leverages implicit feature space alignment and aggregation to model temporal consistency and exploit complementary information from adjacent frames. On the data front, we introduce the first Synthetic Video Bokeh (SVB) dataset, synthesizing authentic bokeh effects using ray-tracing techniques. Furthermore, to improve the robustness of the model to inaccurate disparity maps, we employ a set of augmentation strategies to simulate corrupted disparity inputs during training. Experimental results on both synthetic and real-world data demonstrate the effectiveness of our method.

Depth refinement aims to infer high-resolution depth with fine-grained edges and details, refining low-resolution results of depth estimation models. The prevailing methods adopt tile-based manners by merging numerous patches, which lacks efficiency and produces inconsistency. Besides, prior arts suffer from fuzzy depth boundaries and limited generalizability. Analyzing the fundamental reasons for these limitations, we model depth refinement as a noisy Poisson fusion problem with local inconsistency and edge deformation noises. We propose the Self-distilled Depth Refinement (SDDR) framework to enforce robustness against the noises, which mainly consists of depth edge representation and edge-based guidance. With noisy depth predictions as input, SDDR generates low-noise depth edge representations as pseudo-labels by coarse-to-fine self-distillation. Edge-based guidance with edge-guided gradient loss and edge-based fusion loss serves as the optimization objective equivalent to Poisson fusion. When depth maps are better refined, the labels also become more noise-free. Our model can acquire strong robustness to the noises, achieving significant improvements in accuracy, edge quality, efficiency, and generalizability on five different benchmarks. Moreover, directly training another model with edge labels produced by SDDR brings improvements, suggesting that our method could help with training robust refinement models in future works.

Large-scale datasets with point-wise semantic and instance labels are crucial to 3D instance segmentation but also expensive. To leverage unlabeled data, previous semi-supervised 3D instance segmentation approaches have explored self-training frameworks, which rely on high-quality pseudo labels for consistency regularization. They intuitively utilize both instance and semantic pseudo labels in a joint learning manner. However, semantic pseudo labels contain numerous noise derived from the imbalanced category distribution and natural confusion of similar but distinct categories, which leads to severe collapses in self-training. Motivated by the observation that 3D instances are non-overlapping and spatially separable, we ask whether we can solely rely on instance consistency regularization for improved semi-supervised segmentation. To this end, we propose a novel self-training network InsTeacher3D to explore and exploit pure instance knowledge from unlabeled data. We first build a parallel base 3D instance segmentation model DKNet, which distinguishes each instance from the others via discriminative instance kernels without reliance on semantic segmentation. Based on DKNet, we further design a novel instance consistency regularization framework to generate and leverage high-quality instance pseudo labels. Experimental results on multiple large-scale datasets show that the InsTeacher3D significantly outperforms prior state-of-the-art semi-supervised approaches.

One key issue within one-shot multispectral instance detection (OMID) is to extract features of strong instance discriminative power, domain adaptation capability, and instance-wise generality. Existing methods generally only rely on visual clues. Comparatively, text is advantageous due to its structured information, high semantics, and low noise. Inspired by recent emergence of large image-text datasets and breakthrough visual-language models, we propose to advance OMID with text's supervision for the first time. To this end, our key idea is to establish the relationship between one-shot multispectral instance with ImageNet class labels via the CLIP model. Particularly, we retrieve, rank, and ensemble the text features of ImageNet labels via instance image feature as query. Then the resulting instance image and text features are realigned and fused to obtain a multimodal feature. Meanwhile, a multispectral contrastive learning approach is proposed to drive multimodal feature learning for OMID. Note that all the procedures are end-to-end trained in a unified network. In this way, the instance discriminative power and domain adaptation capability are facilitated simultaneously. Experiments on two tailored multispectral instance detection datasets verify the effectiveness of our method.

In this work, we pay the first effort to address one-shot 3D action recognition in point cloud sequence, without skeleton information. The main contribution lies in two folders. First, a novel one-shot classification approach that considers the feature distribution of 3D action is proposed. We find that, for different 3D actions their dimensional-wise feature distributions are generally in Gaussian form and similar action categories hold approximate feature distributions. Accordingly, K-nearest base classes’ mean value and covariance matrix information help to form one-shot novel class’s pseudo feature distribution. To alleviate the potential ambiguous problem within nearest neighbor search, we divide the base classes into subsets via C-means clustering to facilitate the similarity measure to novel class. Meanwhile, the feature distribution of base class’s whole set and subsets will be jointly considered for generating novel class’s pseudo feature distribution. Multi-dimensional Gaussian sampling is conducted on the acquired pseudo feature distribution for feature-level data augmentation, to make one-shot novel class “never walk alone” for leveraging classifier training. Secondly to better characterize fine-grained 3D action, a temporal attention method is proposed, via introducing vision Transformer (ViT) to capture action’s discriminative short-term motion pattern with densely sampled short-term 3DV (3D dynamic voxel) features along temporal dimension. Experiments on NTU RGB+D 120 and 60 verify superiority of our approach. It outperforms state-of-the-art skeleton-based methods by 13.9% at most. The source code is available athttps://github.com/Tong-XY/YNWA.

Talking head animation transforms a source anime image to a target pose, where the transformation includes the change of facial expression and head movement. In contrast to existing approaches that operate on the low-resolution image (256 × 256), we study this task at a higher resolution,e.g., 512 × 512. High-resolution talking head animation, however, raises two major challenges: i) how to achieve smooth global transformation while maintaining rich details of anime characters under large-displacement pose variations; ii) how to address the shortage of data, because no related dataset is publicly available. In this paper, we present a Hierarchical Feature Warping and Blending (HFWB) model, which tackles talking head animation hierarchically. Specifically, we use low-level features to control global transformation and high-level features to determine the details of anime characters, under the guidance of feature flow fields. These features are then blended by selective fusion units, outputting transformed anime images. In addition, we construct an anime pose dataset–AniTalk-2K, aiming to alleviate the shortage of data. It contains around 2000 anime characters with thousands of different face/head poses at a resolution of 512 × 512. Extensive experiments on AniTalk-2K demonstrate the superiority of our approach in generating high-quality anime talking heads over state-of-the-art methods.

In this paper, a novel LiDAR-Camera Alignment (LCA) method using homogeneous local-global spatial aware representation is proposed. Compared with the state-of-the-art methods (e.g., LCCNet), our proposition holds 2 main superiorities. First, homogeneous multi-modality representation learned with a uniform CNN model is applied along the iterative prediction stages, instead of the state-of-the-art heterogeneous counterparts extracted from the separated modality-wise CNN models within each stage. In this way, the model size can be significantly decreased (e.g., 12.39M (ours) vs. 333.75M (LCCNet)). Meanwhile, within our proposition the interaction between LiDAR and camera data is built during feature learning to better exploit the descriptive clues, which has not been well concerned by the existing approaches. Secondly, we propose to equip the learned LCA representation with local-global spatial aware capacity via encoding CNN’s local convolutional features with Transformer’s non-local self-attention manner. Accordingly, the local fine details and global spatial context can be jointly captured by the encoded local features. And, they will be jointly used for LCA. On the other hand, the existing methods generally choose to reveal the global spatial property via intuitively concatenating the local features. Additionally at the initial LCA stage, LiDAR is roughly aligned with camera by our pre-alignment method, according to the point distribution characteristics of its 2D projection version with the initial extrinsic parameters. Although its structure is simple, it can essentially alleviate LCA’s difficulty for the consequent stages. To better optimize LCA, a novel loss function that builds the correlation between translation and rotation loss items is also proposed. The experiments on KITTI data verifies the superiority of our proposition both on effectiveness and efficiency. The source code will be released athttps://github.com/Zaf233/Light-weight-LCAupon acceptance.

Cropping box regression algorithms re-frame the images with predicted cropping boxes for better composition quality, which can save considerable manpower and time for massive image retouching work. Yet, recent learning-based cropping box regression algorithms require expert annotations, which makes the scale of training limited. This consequently incurs a performance bottleneck. To address this issue, previous works seek the help from auxiliary datasets of related tasks,e.g., the composition classification. However, the domain gap between related tasks and the likewise restricted scale of auxiliary datasets are still limiting factors. Hence, our work provides a novel semi-supervised framework that can learn better re-framing knowledge with unlimited unlabeled data. We make use of the unlabeled data via pseudo-labeling, where the model learns from the pseudo labels generated from a temporal ensemble version of itself. To prevent the model learns from its own mistakes,a.k.a. the problem of confirmation bias, we propose to rectify the mistakes by fusing multiple candidate pseudo labels into the better ones. The fusion procedure is based on the uncertainty estimation for each boundary of the candidate cropping boxes. The multiple candidates are from the proposed aesthetic region proposal network. Extensive experimental results explain how the uncertainty-based pseudo label fusion procedure overcomes the confirmation bias and demonstrate the superiority of our semi-supervised cropping box regression framework.

Depth information plays a pivotal role in numerous computer vision applications, including autonomous driving, 3D reconstruction, and 3D content generation. When deploying depth estimation models in practical applications, it is essential to ensure that the models have strong generalization capabilities. However, existing depth estimation methods primarily concentrate on robust single-image depth estimation, leading to the occurrence of flickering artifacts when applied to video inputs. On the other hand, video depth estimation methods either consume excessive computational resources or lack robustness. To address the above issues, we propose ViTA, a video transformer adaptor, to estimate temporally consistent video depth in the wild. In particular, we leverage a pre-trained image transformer (i.e., DPT) and introduce additional temporal embeddings in the transformer blocks. Such designs enable our ViTA to output reliable results given an unconstrained video. Besides, we present a spatio-temporal consistency loss for supervision. The spatial loss computes the per-pixel discrepancy between the prediction and the ground truth in space, while the temporal loss regularizes the inconsistent outputs of the same point in consecutive frames. To find the correspondences between consecutive frames, we design a bi-directional warping strategy based on the forward and backward optical flow. During inference, our ViTA no longer requires optical flow estimation, which enables it to estimate spatially accurate and temporally consistent video depth maps with fine-grained details in real time. We conduct a detailed ablation study to verify the effectiveness of the proposed components. Extensive experiments on the zero-shot cross-dataset evaluation demonstrate that the proposed method is superior to previous methods.

We show that crowd counting can be formulated as a sequential decision-making (SDM) problem. Inspired by human counting, we evade one-step estimation mostly executed in existing counting models and decompose counting into sequential sub-decision problems. During implementation, a key insight is to interpret sequential counting as a physical process in reality-scale weighing. This analogy allows us to implement a novel "counting scale" termed LibraNet. Our idea is that, by placing a crowd image on the scale, LibraNet (agent) learns to place appropriate weights to match the count: at each step, one weight (action) is chosen from the weight box (the predefined action pool) conditioned on the image features and the placed weights (state) until the pointer (the agent output) informs balance. We investigate two forms of state definition and explore four types of LibraNet implementations under different learning paradigms, including deep Q-network (DQN), actor-critic (AC), imitation learning (IL), and mixed AC+IL. Experiments show that LibraNet indeed mimics scale weighing, that it outperforms or performs comparably against state-of-the-art approaches on five crowd counting benchmarks, that it can be used as a plug-in to improve off-the-shelf counting models, and particularly that it demonstrates remarkable cross-dataset generalization. Code and models are available at https://git.io/libranet.

This work pays the first research effort to address unsupervised 3-D action representation learning with point cloud sequence, which is different from existing unsupervised methods that rely on 3-D skeleton information. Our proposition is built on the state-of-the-art 3-D action descriptor 3-D dynamic voxel (3DV) with contrastive learning (CL). The 3DV can compress the point cloud sequence into a compact point cloud of 3-D motion information. Spatiotemporal data augmentations are conducted on it to drive CL. However, we find that existing CL methods (e.g., SimCLR or MoCo v2) often suffer from high pattern variance toward the augmented 3DV samples from the same action instance, that is, the augmented 3DV samples are still of high feature complementarity after CL, while the complementary discriminative clues within them have not been well exploited yet. To address this, a feature augmentation adapted CL (FACL) approach is proposed, which facilitates 3-D action representation via concerning the features from all augmented 3DV samples jointly, in spirit of feature augmentation. FACL runs in a global-local way: one branch learns global feature that involves the discriminative clues from the raw and augmented 3DV samples, and the other focuses on enhancing the discriminative power of local feature learned from each augmented 3DV sample. The global and local features are fused to characterize 3-D action jointly via concatenation. To fit FACL, a series of spatiotemporal data augmentation approaches is also studied on 3DV. Wide-range experiments verify the superiority of our unsupervised learning method for 3-D action feature learning. It outperforms the state-of-the-art skeleton-based counterparts by 6.4% and 3.6% with the cross-setup and cross-subject test settings on NTU RGB+D 120, respectively. The source code is available at https://github.com/tangent-T/FACL.

Automatic image cropping algorithms aim to recompose images like human-being photographers by generating the cropping boxes with improved composition quality. Cropping box regression approaches learn the beauty of composition from annotated cropping boxes. However, the bias of annotations leads to quasi-trivial recomposing results, which has an obvious tendency to the average location of training samples. The crux of this predicament is that the task is naively treated as a box regression problem, where rare samples might be dominated by normal samples, and the composition patterns of rare samples are not well exploited. Observing that similar composition patterns tend to be shared by the cropping boundaries annotated nearly, we argue to find the beauty of composition from the rare samples by clustering the samples with similar cropping boundary annotations, i.e., similar composition patterns. We propose a novel Contrastive Composition Clustering (C2C) to regularize the composition features by contrasting dynamically established similar and dissimilar pairs. In this way, common composition patterns of multiple images can be better summarized, which especially benefits the rare samples and endows our model with better generalizability to render nontrivial results. Extensive experimental results show the superiority of our model compared with prior arts. We also illustrate the philosophy of our design with an interesting analytical visualization.

We study the composition style in deep image matting, a notion that characterizes a data generation flow on how to exploit limited foregrounds and random backgrounds to form a training dataset. Prior art executes this flow in a completely random manner by simply going through the foreground pool or by optionally combining two foregrounds before foreground-background composition. In this work, we first show that naive foreground combination can be problematic and therefore derive an alternative formulation to reasonably combine foregrounds. Our second contribution is an observation that matting performance can benefit from a certain occurrence frequency of combined foregrounds and their associated source foregrounds during training. Inspired by this, we introduce a novel composition style that binds the source and combined foregrounds in a definite triplet. In addition, we also find that different orders of foreground combination lead to different foreground patterns, which further inspires a quadruplet-based composition style. Results under controlled experiments on four matting baselines show that our composition styles outperform existing ones and invite consistent performance improvement on both composited and real-world datasets. Code is available at: https://github.com/coconuthust/composition_styles

Correspondence pruning aims to search consistent correspondences (inliers) from a set of putative correspondences. It is challenging because of the disorganized spatial distribution of numerous outliers, especially when putative correspondences are largely dominated by outliers. It's more challenging to ensure effectiveness while maintaining efficiency. In this paper, we propose an effective and efficient method for correspondence pruning. Inspired by the success of attentive context in correspondence problems, we first extend the attentive context to the first-order attentive context and then introduce the idea of attention in attention (ANA) to model second-order attentive context for correspondence pruning. Compared with first-order attention that focuses on feature-consistent context, second-order attention dedicates to attention weights itself and provides an additional source to encode consistent context from the attention map. For efficiency, we derive two approximate formulations for the naive implementation of second-order attention to optimize the cubic complexity to linear complexity, such that second-order attention can be used with negligible computational overheads. We further implement our formulations in a second-order context layer and then incorporate the layer in an ANA block. Extensive experiments demonstrate that our method is effective and efficient in pruning outliers, especially in high-outlier-ratio cases. Compared with the state-of-the-art correspondence pruning approach LMCNet, our method runs 14 times faster while maintaining a competitive accuracy.

Category-agnostic pose estimation (CAPE) aims to predict keypoints for arbitrary categories given support images with keypoint annotations. Existing approaches match the keypoints across the image for localization. However, such a one-stage matching paradigm shows inferior accuracy: the prediction heavily relies on the matching results, which can be noisy due to the open set nature in CAPE. For example, two mirror-symmetric keypoints (e.g., left and right eyes) in the query image can both trigger high similarity on certain support keypoints (eyes), which leads to duplicated or opposite predictions. To calibrate the inaccurate matching results, we introduce a two-stage framework, where matched keypoints from the first stage are viewed as similarity-aware position proposals. Then, the model learns to fetch relevant features to correct the initial proposals in the second stage. We instantiate the framework with a transformer model tailored for CAPE. The transformer encoder incorporates specific designs to improve the representation and similarity modeling in the first matching stage. In the second stage, similarity-aware proposals are packed as queries in the decoder for refinement via cross-attention. Our method surpasses the previous best approach by large margins on CAPE benchmark MP-100 on both accuracy and efficiency. Code available at github.com/flyinglynx/CapeFormer

3D interacting hand pose estimation from a single RGB image is a challenging task, due to serious self-occlusion and inter-occlusion towards hands, confusing similar appearance patterns between 2 hands, ill-posed joint position mapping from 2D to 3D, etc.. To address these, we propose to extend A2J-the state-of-the-art depth-based 3D single hand pose estimation method-to RGB domain under interacting hand condition. Our key idea is to equip A2J with strong local-global aware ability to well capture interacting hands' local fine details and global articulated clues among joints jointly. To this end, A2J is evolved under Transformer's non-local encoding-decoding framework to build A2J- Transformer. It holds 3 main advantages over A2J. First, self-attention across local anchor points is built to make them global spatial context aware to better capture joints' articulation clues for resisting occlusion. Secondly, each anchor point is regarded as learnable query with adaptive feature learning for facilitating pattern fitting capacity, instead of having the same local representation with the others. Last but not least, anchor point locates in 3D space instead of 2D as in A2J, to leverage 3D pose prediction. Experiments on challenging InterHand 2.6M demonstrate that, A2J-Transformer can achieve state-of-the-art model-free performance (3.38mm MPJPE advancement in 2-hand case) and can also be applied to depth domain with strong generalization. The code is avaliable at https://github.com/ChanglongJiangGit/A2J-Transformer.

We present 3D Cinemagraphy, a new technique that mar-ries 2D image animation with 3D photography. Given a single still image as input, our goal is to generate a video that contains both visual content animation and camera motion. We empirically find that naively combining existing 2D image animation and 3D photography methods leads to obvious artifacts or inconsistent animation. Our key insight is that representing and animating the scene in 3D space offers a natural solution to this task. To this end, we first convert the input image into feature-based layered depth images using predicted depth values, followed by unprojecting them to a feature point cloud. To animate the scene, we perform motion estimation and lift the 2D motion into the 3D scene flow. Finally, to resolve the problem of hole emer-gence as points move forward, we propose to bidirectionally displace the point cloud as per the scene flow and synthe-size novel views by separately projecting them into target image planes and blending the results. Extensive experiments demonstrate the effectiveness of our method. A user study is also conducted to validate the compelling rendering results of our method.

Real-time eyeblink detection in the wild can widely serve for fatigue detection, face anti-spoofing, emotion analysis, etc. The existing research efforts generally focus on single-person cases towards trimmed video. However, multi-person scenario within untrimmed videos is also important for practical applications, which has not been well concerned yet. To address this, we shed light on this research field for the first time with essential contributions on dataset, theory, and practices. In particular, a large-scale dataset termed MPEblink that involves 686 untrimmed videos with 8748 eyeblink events is proposed under multi-person conditions. The samples are captured from uncon-strainedfilms to reveal “in the wild“ characteristics. Meanwhile, a real-time multi-person eyeblink detection method is also proposed. Being different from the existing counter-parts, our proposition runs in a one-stage spatio-temporal way with end-to-end learning capacity. Specifically, it simultaneously addresses the sub-tasks of face detection, face tracking, and human instance-level eyeblink detection. This paradigm holds 2 main advantages: (1) eyeblink features can be facilitated via the face's global context (e.g., head pose and illumination condition) with joint optimization and interaction, and (2) addressing these sub-tasks in parallel instead of sequential manner can save time remarkably to meet the real-time running requirement. Experiments on MPEblink verify the essential challenges of real-time multi-person eyeblink detection in the wild for untrimmed video. Our method also outperforms existing approaches by large margins and with a high inference speed.

Video depth estimation aims to infer temporally consistent depth. Some methods achieve temporal consistency by finetuning a single-image depth model during test time using geometry and re-projection constraints, which is inefficient and not robust. An alternative approach is to learn how to enforce temporal consistency from data, but this requires well-designed models and sufficient video depth data. To address these challenges, we propose a plug-and-play framework called Neural Video Depth Stabilizer (NVDS) that stabilizes inconsistent depth estimations and can be applied to different single-image depth models without extra effort. We also introduce a large-scale dataset, Video Depth in the Wild (VDW), which consists of 14,203 videos with over two million frames, making it the largest natural-scene video depth dataset to our knowledge. We evaluate our method on the VDW dataset as well as two public benchmarks and demonstrate significant improvements in consistency, accuracy, and efficiency compared to previous approaches. Our work serves as a solid baseline and provides a data foundation for learning-based video depth models. We will release our dataset and code for future research.

We present DySample, an ultra-lightweight and effective dynamic upsampler. While impressive performance gains have been witnessed from recent kernel-based dynamic upsamplers such as CARAFE, FADE, and SAPA, they introduce much workload, mostly due to the time-consuming dynamic convolution and the additional sub-network used to generate dynamic kernels. Further, the need for high-res feature guidance of FADE and SAPA somehow limits their application scenarios. To address these concerns, we bypass dynamic convolution and formulate upsampling from the perspective of point sampling, which is more resource-efficient and can be easily implemented with the standard built-in function in PyTorch. We first showcase a naive design, and then demonstrate how to strengthen its upsampling behavior step by step towards our new upsampler, DySample. Compared with former kernel-based dynamic upsamplers, DySample requires no customized CUDA package and has much fewer parameters, FLOPs, GPU memory, and latency. Besides the light-weight characteristics, DySample outperforms other upsamplers across five dense prediction tasks, including semantic segmentation, object detection, instance segmentation, panoptic segmentation, and monocular depth estimation. Code is available at https://github.com/tiny-smart/dysample.

We show that crowd counting can be viewed as a decomposable point querying process. This formulation enables arbitrary points as input and jointly reasons whether the points are crowd and where they locate. The querying processing, however, raises an underlying problem on the number of necessary querying points. Too few imply underestimation; too many increase computational overhead. To address this dilemma, we introduce a decomposable structure, i.e., the point-query quadtree, and propose a new counting model, termed Point quEry Transformer (PET). PET implements decomposable point querying via data-dependent quadtree splitting, where each querying point could split into four new points when necessary, thus enabling dynamic processing of sparse and dense regions. Such a querying process yields an intuitive, universal modeling of crowd as both the input and output are interpretable and steerable. We demonstrate the applications of PET on a number of crowd-related tasks, including fully-supervised crowd counting and localization, partial annotation learning, and point annotation refinement, and also report state-of-the-art performance. For the first time, we show that a single counting model can address multiple crowd-related tasks across different learning paradigms. Code is available at https://github.com/cxliu0/PET.

Learning-based multi-view stereo (MVS) method heavily relies on feature matching, which requires distinctive and descriptive representations. An effective solution is to apply non-local feature aggregation, e.g., Transformer. Albeit useful, these techniques introduce heavy computation overheads for MVS. Each pixel densely attends to the whole image. In contrast, we propose to constrain nonlocal feature augmentation within a pair of lines: each point only attends the corresponding pair of epipolar lines. Our idea takes inspiration from the classic epipolar geometry, which shows that one point with different depth hypotheses will be projected to the epipolar line on the other view. This constraint reduces the 2D search space into the epipolar line in stereo matching. Similarly, this suggests that the matching of MVS is to distinguish a series of points lying on the same line. Inspired by this point-toline search, we devise a line-to-point non-local augmentation strategy. We first devise an optimized searching algorithm to split the 2D feature maps into epipolar line pairs. Then, an Epipolar Transformer (ET) performs non-local feature augmentation among epipolar line pairs. We incorporate the ET into a learning-based MVS baseline, named ET-MVSNet. ET-MVSNet achieves state-of-the-art reconstruction performance on both the DTU and Tanks-and-Temples benchmark with high efficiency. Code is available at https://github.com/TQTQliu/ET-MVSNet.

Learning-based multi-view stereo (MVS) methods deal with predicting accurate depth maps to achieve an accurate and complete 3D representation. Despite the excellent performance, existing methods ignore the fact that a suitable depth geometry is also critical in MVS. In this paper, we demonstrate that different depth geometries have significant performance gaps, even using the same depth prediction error. Therefore, we introduce an ideal depth geometry composed of Saddle-Shaped Cells, whose predicted depth map oscillates upward and downward around the ground-truth surface, rather than maintaining a continuous and smooth depth plane. To achieve it, we develop a coarse-to-fine framework called Dual-MVSNet (DMVSNet), which can produce an oscillating depth plane. Technically, we predict two depth values for each pixel (Dual-Depth), and propose a novel loss function and a checkerboard-shaped selecting strategy to constrain the predicted depth geometry. Compared to existing methods, DMVSNet achieves a high rank on the DTU benchmark and obtains the top performance on challenging scenes of Tanks and Temples, demonstrating its strong performance and generalization ability. Our method also points to a new research direction for considering depth geometry in MVS.

Video stabilization refers to the problem of transforming a shaky video into a visually pleasing one. The question of how to strike a good trade-off between visual quality and computational speed has remained one of the open challenges in video stabilization. Inspired by the analogy between wobbly frames and jigsaw puzzles, we propose an iterative optimization-based learning approach using synthetic datasets for video stabilization, which consists of two interacting submodules: motion trajectory smoothing and full-frame outpainting. First, we develop a two-level (coarse-to-fine) stabilizing algorithm based on the probabilistic flow field. The confidence map associated with the estimated optical flow is exploited to guide the search for shared regions through backpropagation. Second, we take a divide-and-conquer approach and propose a novel multi-frame fusion strategy to render full-frame stabilized views. An important new insight brought about by our iterative optimization approach is that the target video can be interpreted as the fixed point of nonlinear mapping for video stabilization. We formulate video stabilization as a problem of minimizing the amount of jerkiness in motion trajectories, which guarantees convergence with the help of fixed-point theory. Extensive experimental results are reported to demonstrate the superiority of the proposed approach in terms of computational speed and visual quality. The code will be available on GitHub.

Human Mesh Reconstruction (HMR) aims to recover 3D human poses and shapes from a single image. Existing parameterized HMR approaches follow the "representation-to-reasoning'' paradigm to predict human body and pose parameters. This paradigm typically involves intermediate representation and complex pipeline, where potential side effects may occur that could hinder performance. In contrast, query-based non-parameterized methods directly output 3D joints and mesh vertices, but they rely on excessive queries for prediction, leading to low efficiency and robustness. In this work, we propose a simple query-based framework, dubbed SimHMR, for parameterized human mesh reconstruction. This framework streamlines the prediction process by using a few parameterized queries, which effectively removes the need for hand-crafted intermediate representation and reasoning pipeline. Different from query-based non-parameterized HMR that uses excessive coordinate queries, SimHMR only requires a few semantic queries, which physically correspond to pose, shape, and camera. The use of semantic queries significantly improves the efficiency and robustness in extreme scenarios, e.g., occlusions. Without bells and whistles, øurs achieves state-of-the-art performance on 3DPW and Human3.6M benchmarks, and surpasses existing methods on challenging 3DPW-OCC. Code available at https://github.com/inso-13/SimHMR github.com/inso-13/SimHMR

Depth estimation aims to predict dense depth maps. In autonomous driving scenes, sparsity of annotations makes the task challenging. Supervised models produce concave objects due to insufficient structural information. They overfit to valid pixels and fail to restore spatial structures. Self-supervised methods are proposed for the problem. Their robustness is limited by pose estimation, leading to erroneous results in natural scenes. In this paper, we propose a supervised framework termed Diffusion-Augmented Depth Prediction (DADP). We leverage the structural characteristics of diffusion model to enforce depth structures of depth models in a plug-and-play manner. An object-guided integrality loss is also proposed to further enhance regional structure integrality by fetching objective information. We evaluate DADP on three driving benchmarks and achieve significant improvements in depth structures and robustness. Our work provides a new perspective on depth estimation with sparse annotations in autonomous driving scenes.

We study the problem of synthesizing a long-term dynamic video from only a single image. This is challenging since it requires consistent visual content movements given large camera motions. Existing methods either hallucinate inconsistent perpetual views or struggle with long camera trajectories. To address these issues, it is essential to estimate the underlying 4D (including 3D geometry and scene motion) and fill in the occluded regions. To this end, we present Make-It-4D, a novel method that can generate a consistent long-term dynamic video from a single image. On the one hand, we utilize layered depth images (LDIs) to represent a scene, and they are then unprojected to form a feature point cloud. To animate the visual content, the feature point cloud is displaced based on the scene flow derived from motion estimation and the corresponding camera pose. Such 4D representation enables our method to maintain the global consistency of the generated dynamic video. On the other hand, we fill in the occluded regions by using a pre-trained diffusion model to inpaint and outpaint the input image. This enables our method to work under large camera motions. Benefiting from our design, our method can be training-free which saves a significant amount of training time. Experimental results demonstrate the effectiveness of our approach, which showcases compelling rendering results.

We introduce Probabilistic Coordinate Fields (PCFs), a novel geometric-invariant coordinate representation for image correspondence problems. In contrast to standard Cartesian coordinates, PCFs encode coordinates in correspondence-specific barycentric coordinate systems (BCS) with affine invariance. To know when and where to trust the encoded coordinates, we implement PCFs in a probabilistic network termed PCF-Net, which parameterizes the distribution of coordinate fields as Gaussian mixture models. By jointly optimizing coordinate fields and their confidence conditioned on dense flows, PCF-Net can work with various feature descriptors when quantifying the reliability of PCFs by confidence maps. An interesting observation of this work is that the learned confidence map converges to geometrically coherent and semantically consistent regions, which facilitates robust coordinate representation. By delivering the confident coordinates to keypoint/feature descriptors, we show that PCF-Net can be used as a plug-in to existing correspondence-dependent approaches. Extensive experiments on both indoor and outdoor datasets suggest that accurate geometric invariant coordinates help to achieve the state of the art in several correspondence problems, such as sparse feature matching, dense image registration, camera pose estimation, and consistency filtering. Further, the interpretable confidence map predicted by PCF-Net can also be leveraged to other novel applications from texture transfer to multi-homography classification.

In this letter, we propose a new method, Multi-Clue Gaze (MCGaze), to facilitate video gaze estimation via capturing spatial-temporal interaction context among head, face, and eye in an end-to-end learning way, which has not been well concerned yet. The main advantage of MCGaze is that the tasks of clue localization of head, face, and eye can be solved jointly for gaze estimation in a one-step way, with joint optimization to seek optimal performance. During this, spatial-temporal context exchange happens among the clues on the head, face, and eye. Accordingly, the final gazes obtained by fusing features from various queries can be aware of global clues from heads and faces, and local clues from eyes simultaneously, which essentially leverages performance. Meanwhile, the one-step running way also ensures high running efficiency. Experiments on the challenging Gaze360 dataset verify the superiority of our proposition. The source code will be released athttps://github.com/zgchen33/MCGaze.

All-in-Focus (AIF) photography is expected to be a commercial selling point for modern smartphones. Standard AIF synthesis requires manual, time-consuming operations such as focal stack compositing, which is unfriendly to ordinary people. To achieve point-and-shoot AIF photography with a smartphone, we expect that an AIF photo can be generated from one shot of the scene, instead of from multiple photos captured by the same camera. Benefiting from the multi-camera module in modern smartphones, we introduce a new task of AIF synthesis from main (wide) and ultra-wide cameras. The goal is to recover sharp details from defocused regions in the main-camera photo with the help of the ultra-wide-camera one. The camera setting poses new challenges such as parallax-induced occlusions and inconsistent color between cameras. To overcome the challenges, we introduce a predict-and-refine network to mitigate occlusions and propose dynamic frequency-domain alignment for color correction. To enable effective training and evaluation, we also build an AIF dataset with 2686 unique scenes. Each scene includes two photos captured by the main camera, one photo captured by the ultra-wide camera, and a synthesized AIF photo. Results show that our solution, termed EasyAIF, can produce high-quality AIF photos and outperforms strong baselines quantitatively and qualitatively. For the first time, we demonstrate point-and-shoot AIF photo synthesis successfully from main and ultra-wide cameras.

LiDAR-Camera alignment (LCA) is an important preprocessing procedure for fusing LiDAR and camera data. For it, one key issue is to extract unified cross-modality representation for characterizing the heterogeneous LiDAR and camera data effectively and robustly. The main challenge is to resist the modality gap and visual data degradation during feature learning, while still maintaining strong representative power. To address this, a novel modality adapted local-to-global representation learning method is proposed. The research efforts are paid in 2 main folders via modality adaptation and capturing global spatial context. First for modality gap resistance, LiDAR and camera data is projected into the same depth map domain for unified representation learning. Particularly, LiDAR data is converted to depth map according to pre-acquired extrinsic parameters. Thanks to the recent advantage of deep learning based monocular depth estimation, camera data is transformed into depth map in data driven manner, which is jointly optimized with LCA. Secondly to capture global spatial context, ViT (vision transformer) is introduced to LCA. The concept of LCA token is proposed for aggregating the local spatial patterns to form global spatial representation with transformer encoding. And, it is shared by all the samples. In this way, it can involve global sample-level information to leverage generalization ability. The experiments on KITTI dataset verify superiority of our proposition. Furthermore, the proposed approach is more robust to camera data degeneration (e.g., imaging blurring and noise) often faced by the practical applications. Under some challenging test cases, the performance advancement of our method is over$1.9~cm$/4.1° on translation / rotation error. While our model size (8.77M) is much smaller than existing methods (e.g., LCCNet of 66.75M). The source code will be released athttps://github.com/Zaf233/RLCAupon acceptance.

Most current infrared small target detection methods attempt to fuse local and global information by using single-scale inputs and creating a multi-scale feature pyramid during network feeding forwards. Further to this, our research finds that using high-resolution inputs can improve recall, while low-resolution inputs improve precision. Nevertheless, solely focusing on global or local information can result in missing target and false alarm. To address these issues, we propose the BPR-Net to balance precision and recall via a novel multi-scale attention mechanism, which combines semantic and shallow features of multi-scale inputs. We first scale the input image into multiple images with varying resolutions and feed them into the network. In the encoder, Scale Fusion Module (SFM) fuses features from corresponding images of different resolutions. In the decoder, a Channel Fusion Module (CFM) fuses useful information from multiple channels. Furthermore, a Wavelet Transform cross-layer skip Layer (WTL) is employed to enhance the interaction between decoder layers for more effective multi-scale feature fusion. Experimental results demonstrate that our approach achieves a balance between recall and precision and yields state-of-the-art performance on challenging benchmarks including Sirst, MDvsFA, and SIATD. Notably, our approach achieves an F1 score of 0.9409 on the challenging benchmark SIATD, surpassing the state-of-the-art method by 16.7%.

Real-time eyeblink detection in the wild is a recently emerged challenging task that suffers from dramatic variations in face attribute, pose, illumination, camera view and distance, etc. One key issue is to well characterize eyelid’s intrinsic motion (i.e., approaching and departure between upper and lower eyelid) robustly, under unconstrained conditions. Towards this, a novel eyelid’s intrinsic motion-aware feature learning approach is proposed. Our proposition lies in 3 folds. First, the feature extractor is led to focus on informative eye region adaptively via introducing visual attention in a coarse-to-fine way, to guarantee robustness and fine-grained descriptive ability jointly. Then, 2 constraints are proposed to make feature learning be aware of eyelid’s intrinsic motion. Particularly, one concerns the fact that the inter-frame feature divergence within eyeblink processes should be greater than non-eyeblink ones to better reveal eyelid’s intrinsic motion. The other constraint minimizes the inter-frame feature divergence of non-eyeblink samples, to suppress motion clues due to head or camera movement, illumination change, etc. Meanwhile, concerning the high ambiguity between eyeblink and non-eyeblink samples, soft sample labels are acquired via self-knowledge distillation to conduct feature learning with finer supervision than the hard ones. The experiments verify that, our proposition is significantly superior to the state-of-the-art ones (i.e., advantage on F1-score over 7%) and with real-time running efficiency. It is also of strong generalization capacity towards constrained conditions. The source code is available athttps://github.com/wenzhengzeng/blink_eyelid.

We propose BokehMe, a hybrid bokeh rendering framework that marries a neural renderer with a classical physically motivated renderer. Given a single image and a potentially imperfect disparity map, BokehMe generates high-resolution photo-realistic bokeh effects with adjustable blur size, focal plane, and aperture shape. To this end, we analyze the errors from the classical scattering-based method and derive a formulation to calculate an error map. Based on this formulation, we implement the classical renderer by a scattering-based method and propose a two-stage neural renderer to fix the erroneous areas from the classical renderer. The neural renderer employs a dynamic multi-scale scheme to efficiently handle arbitrary blur sizes, and it is trained to handle imperfect disparity input. Experiments show that our method compares favorably against previous methods on both synthetic image data and real image data with predicted disparity. A user study is further conducted to validate the advantage of our method.

Class-agnostic counting (CAC) aims to count all instances in a query image given few exemplars. A standard pipeline is to extract visual features from exemplars and match them with query images to infer object counts. Two essential components in this pipeline are feature representation and similarity metric. Existing methods either adopt a pretrained network to represent features or learn a new one, while applying a naive similarity metric with fixed inner product. We find this paradigm leads to noisy similarity matching and hence harms counting performance. In this work, we propose a similarity-aware CAC framework that jointly learns representation and similarity metric. We first instantiate our framework with a naive baseline called Bilinear Matching Network (BMNet), whose key component is a learnable bilinear similarity metric. To further embody the core of our framework, we extend BMNet to BMNet+ that models similarity from three aspects: 1) representing the instances via their self-similarity to enhance feature robustness against intra-class variations; 2) comparing the similarity dynamically to focus on the key patterns of each exemplar; 3) learning from a supervision signal to impose explicit constraints on matching results. Extensive experiments on a recent CAC dataset FSC147 show that our models significantly outperform state-of-the-art CAC approaches. In addition, we also validate the cross-dataset generality of BMNet and BMNet+ on a car counting dataset CARPK. Code is at tiny.one/BMNet

Image translation from human faces to anime ones brings a low-end, efficient way to create animation characters for animation industry. However, due to the significant inter-domain difference between anime images and human photos, existing image-to-image translation approaches cannot address this task well. To solve this dilemma, we propose HyProGAN, an exemplar-guided image-to-image translation model without paired data. The key contribution of HyPro-GAN is that it introduces a novel hybrid and progressive training strategy that expands the unidirectional translation between two domains into the bidirectional intra-domain and inter-domain translation. To enhance the consistency between input and output, we further propose a local masking loss to align the facial features between the human face and the generated anime face. Extensive experiments demonstrate the superiority of HyProGAN against state-of-the-art models.

Image cropping aims to enhance the aesthetic quality of a given image by searching for the good cropping views. One common routine is to score and rank the candidate views by the neural network. The network is expected to discriminate the subtle view-wise differences. However, the image-wise differences and the ambiguity in the annotations render difficulties in discriminating the view-wise differences. To focus on the view-wise differences, we propose a feature spliter to build image-wise and view-wise feature and evaluate the candidate views only based on the view-wise feature. Then, we propose the ranking gain loss that alleviates the ambiguity in annotations to amplify the view-wise differences. The remarkable improvement compared with prior arts on public benchmarks illustrates that the view-wise differences matter in cropping view recommendation.

Generative adversarial networks,(GANs) have been trained to be professional artists able to create stunning artworks such as face generation and image style transfer. In this paper, we focus on a realistic business scenario: automated generation of customizable icons given desired mobile applications and theme styles. We first introduce a theme-application icon dataset, termed AppIcon, where each icon has two orthogonal theme and app labels. By investigating a strong baseline StyleGAN2, we observe mode collapse caused by the entanglement of the orthogonal labels. To solve this challenge, we propose IconGAN composed of a conditional generator and dual discriminators with orthogonal augmentations, and a contrastive feature disentanglement strategy is further designed to regularize the feature space of the two discriminators. Compared with other approaches, IconGAN indicates a superior advantage on the AppIcon benchmark. Further analysis also justifies the effectiveness of disentangling app and theme representations. Our project will be released at: https://github.com/architect-road/IconGAN.

Temporal consistency is the key challenge of video depth estimation. Previous works are based on additional optical flow or camera poses, which is time-consuming. By contrast, we derive consistency with less information. Since videos inherently exist with heavy temporal redundancy, a missing frame could be recovered from neighboring ones. Inspired by this, we propose the frame masking network (FMNet), a spatial-temporal transformer network predicting the depth of masked frames based on their neighboring frames. By reconstructing masked temporal features, the FMNet can learn intrinsic inter-frame correlations, which leads to consistency. Compared with prior arts, experimental results demonstrate that our approach achieves comparable spatial accuracy and higher temporal consistency without any additional information. Our work provides a new perspective on consistent video depth estimation.

Neural Radiance Field (NeRF) and its variants have exhibited great success on representing 3D scenes and synthesizing photo-realistic novel views. However, they are generally based on the pinhole camera model and assume all-in-focus inputs. This limits their applicability as images captured from the real world often have finite depth-of-field (DoF). To mitigate this issue, we introduce DoF-NeRF, a novel neural rendering approach that can deal with shallow DoF inputs and can simulate DoF effect. In particular, it extends NeRF to simulate the aperture of lens following the principles of geometric optics. Such a physical guarantee allows DoF-NeRF to operate views with different focus configurations. Benefiting from explicit aperture modeling, DoF-NeRF also enables direct manipulation of DoF effect by adjusting virtual aperture and focus parameters. It is plug-and-play and can be inserted into NeRF-based frameworks. Experiments on synthetic and real-world datasets show that, DoF-NeRF not only performs comparably with NeRF in the all-in-focus setting, but also can synthesize all-in-focus novel views conditioned on shallow DoF inputs. An interesting application of DoF-NeRF to DoF rendering is also demonstrated. The source code will be made available at: https://github.com/zijinwuzijin/DoF-NeRF.

We introduce point affiliation into feature upsampling, a notion that describes the affiliation of each upsampled point to a semantic cluster formed by local decoder feature points with semantic similarity. By rethinking point affiliation, we present a generic formulation for generating upsampling kernels. The kernels encourage not only semantic smoothness but also boundary sharpness in the upsampled feature maps. Such properties are particularly useful for some dense prediction tasks such as semantic segmentation. The key idea of our formulation is to generate similarity-aware kernels by comparing the similarity between each encoder feature point and the spatially associated local region of decoder features. In this way, the encoder feature point can function as a cue to inform the semantic cluster of upsampled feature points. To embody the formulation, we further instantiate a lightweight upsampling operator, termed Similarity-Aware Point Affiliation (SAPA), and investigate its variants. SAPA invites consistent performance improvements on a number of dense prediction tasks, including semantic segmentation, object detection, depth estimation, and image matting. Code is available at: https://github.com/poppinace/sapa

Developing accurate algorithms for agricultural pattern recognition from aerial imagery has become increasingly important due to the prevalence of unmanned aerial vehicles (UAVs). This letter introduces a deep encoder–decoder network for semantic land cover segmentation, where the goal is to classify six anomaly categories from multispectral aerial imagery. Since aerial imagery exhibits specific characteristics and visual challenges in this imaging domain, existing semantic segmentation models are not plug-and-play. Starting from a state-of-the-art segmentation model, we present a step-by-step analysis of key challenges and also reveal our observations in addressing these challenges. In particular, we investigate on how to exploit data prior knowledge, how to deal with sample imbalance, and how to encode global semantic and contextual information to improve segmentation. Experiments on a recent large-scale aerial land cover data set demonstrate that our method achieves compelling performance against other state-of-the-art approaches. Our results and insights can provide references for practitioners working in this field when dealing with similar segmentation problems.

Cloud detection and recognition are two important tasks usually referring to image binary segmentation and image-level classification individually. Cloud genera segmentation has more practical significance but is much more challenging as a fine-grained pixel-level dense prediction problem. In this letter, a cascade cloud attribute discrimination network (CCAD-Net) is proposed. Based on an improved encoding-decoding model, CCAD-Net adds a binary segmentation branch for cloud detection and a attribute discrimination branch for cloud attribute feature learning in the decoding stage. Especially, in the attribute discrimination branch, several visual attributes are selected to design the attribute discrimination constraint according to prior professional knowledge and corresponding loss function is defined. These two additional branches and the final cloud genera segmentation branch extract their task-specific features successively and form a cascade structure. Due to the fusion of raw feature, binary segmentation feature, attribute discrimination feature, and cloud genera feature, CCAD-Net can achieve significant better performance than the state-of-the-art methods in cloud genera segmentation in whole-sky images.

This paper focuses on developing efficient and robust evaluation metrics for RANSAC hypotheses to achieve accurate 3D rigid registration. Estimating six-degree-of-freedom (6-DoF) pose from feature correspondences remains a popular approach to 3D rigid registration, where random sample consensus (RANSAC) is a well-known solution to this problem. However, existing metrics for RANSAC hypotheses are either time-consuming or sensitive to common nuisances, parameter variations, and different application scenarios, resulting in performance deterioration with respect to overall registration accuracy and speed. We alleviate this problem by first analyzing the contributions of inliers and outliers and then proposing several efficient and robust metrics with different designing motivations for RANSAC hypotheses. Comparative experiments on four standard datasets with different nuisances and application scenarios verify that our considered metrics can significantly improve the registration performance and are more robust than several state-of-the-art competitors, making them good gifts to practical applications. This work also draws an interesting conclusion, i.e., not all inliers are equal while all outliers should be equal, which may shed new light on this research problem.

The core prerequisite of most modern trackers is a motion assumption, defined as predicting the current location in a limited search region centering at the previous prediction. For clarity, the central subregion of a search region is denoted as the tracking anchor (e.g., the location of the previous prediction in the current frame). However, providing accurate predictions in all frames is very challenging in the complex nature scenes. In addition, the target locations in consecutive frames often change violently under the attribute of fast motion. Both facts are likely to lead the previous prediction to an unbelievable tracking anchor, which will make the aforementioned prerequisite invalid and cause tracking drift. To enhance the reliability of tracking anchors, we propose a real-time multianchor visual tracking mechanism, called multianchor tracking (MAT). Instead of directly relying on the tracking anchor inherited from the previous prediction, MAT selects the best anchor from an anchor ensemble, which includes several objectness-based anchor proposals and the anchor inherited from the previous prediction. The objectness-based anchors provide several complementary selective search regions, and an entropy-minimization-based selection method is introduced to find the best anchor. Our approach offers two benefits: 1) selective search regions can increase the chance of tracking success with affordable computational load and 2) anchor selection introduces the best anchor for each frame, which breaks the limitation of solo depending on the previous prediction. The extensive experiments of nine base trackers upgraded by MAT on four challenging datasets demonstrate the effectiveness of MAT.

Face verification can be regarded as a two-class fine-grained visual-recognition problem. Enhancing the feature's discriminative power is one of the key problems to improve its performance. Metric-learning technology is often applied to address this need while achieving a good tradeoff between underfitting, and overfitting plays a vital role in metric learning. Hence, we propose a novel ensemble cascade metric-learning (ECML) mechanism. In particular, hierarchical metric learning is executed in a cascade way to alleviate underfitting. Meanwhile, at each learning level, the features are split into nonoverlapping groups. Then, metric learning is executed among the feature groups in the ensemble manner to resist overfitting. Considering the feature distribution characteristics of faces, a robust Mahalanobis metric-learning method (RMML) with a closed-form solution is additionally proposed. It can avoid the computation failure issue on an inverse matrix faced by some well-known metric-learning approaches (e.g., KISSME). Embedding RMML into the proposed ECML mechanism, our metric-learning paradigm (EC-RMML) can run in the one-pass learning manner. The experimental results demonstrate that EC-RMML is superior to state-of-the-art metric-learning methods for face verification. The proposed ECML mechanism is also applicable to other metric-learning approaches.

Fast and accurate plant counting tools affect revolution in modern agriculture. Agricultural practitioners, however, expect the output of the tools to be not only accurate but also explainable. Such explainability often refers to the ability to infer which instance is counted. One intuitive way is to generate a bounding box for each instance. Nevertheless, compared with counting by detection, plant counts can be inferred more directly in the local count framework, while one thing reproaching this paradigm is its poor explainability of output visualization. In particular, we find that the poor explainability becomes a bottleneck limiting the counting performance. To address this, we explore the idea of guided upsampling and background suppression where a novel upsampling operator is proposed to allow count redistribution, and segmentation decoders with different fusion strategies are investigated to suppress background, respectively. By integrating them into our previous counting model TasselNetV2, we introduce TasselNetV3 series: TasselNetV3-Lite and TasselNetV3-Seg. We validate the TasselNetV3 series on three public plant counting data sets and a new unmanned aircraft vehicle (UAV)-based data set, covering maize tassels counting, wheat ears counting, and rice plants counting. Extensive results show that guided upsampling and background suppression not only improve counting performance but also enable explainable visualization. Aside from state-of-the-art performance, we have several interesting observations: 1) a limited-receptive-field counter in most cases outperforms a large-receptive-field one; 2) it is sufficient to generate empirical segmentation masks from dotted annotations; 3) middle fusion is a good choice to integrate foreground–backgrounda prioriknowledge; and 4) decoupling the learning of counting and segmentation matters.

Infrared small target detection plays an important role in target warning, ground monitoring, and flight guidance. Existing methods typically utilize local-contrast information of each pixel to detect infrared small targets, neglecting the interior relation between target pixels or background pixels. The mere use of the local information of one pixel, however, is not sufficient for accurate detection, which may lead to missing detection and false alarms. As a harmonious whole, information between pixels are necessary to determine if a pixel belongs to the target or the background. Motivated by the fact that pixels from targets or backgrounds are correlated with each other, we propose a coarse-to-fine interior attention-aware network (IAANet) for infrared small target detection. Specifically, a region proposal network (RPN) is first applied to obtain coarse target regions and filter out backgrounds. Then, we leverage a transformer encoder to model the attention between pixels in coarse target regions, outputting attention-aware features. Finally, predictions are obtained by feeding attention-aware features to a classification head. Extensive experiments show that our approach is capable of detecting targets precisely, of suppressing a variety of false alarm sources, and works effectively in various background environments and target appearances. We show that our IAANet outperforms the state-of-the-art methods by a large margin. Code will be made available at:https://github.com/kwwcv/iaanet.

Practically, person re-identification (re-ID) may suffer from the critical spatial misalignment problem due to inaccurate human detection, variation on human pose and camera viewpoint, etc. To address this, a hierarchical discriminative spatial aggregation method is proposed. The key idea is to conduct spatial aggregation on local human parts via global average-pooling to acquire the strong spatial misalignment tolerance, with VALD encoding on the local parts for facilitating discriminative power jointly. This proposition is built on NetVLAD to ensure end-to-end deep learning capacity. Due to the fine-grained property of person re-ID task that has not been well concerned by the original NetVLAD model for scene recognition, a feature refinement layer that consists of 1 fully-connected (FC) layer and 2 batch normalization (BN) layers is added on top of the raw NetVLAD layer to enhance the discriminative power and training convergence. And, a human body occlusion and background component dropout manner is also proposed to resist the effect of serious occlusion. Technically, a refined codeword initialization manner is proposed to alleviate the potential codeword imbalance problem caused by naive random initialization. The proposed discriminative spatial aggregation approach is then conducted on multi-resolution convolutional feature map layers hierarchically via early feature fusion, to involve richer semantic and fine-grained visual clues jointly. Wide-range experiments on 6 datasets (i.e., CUHK03, DukeMTMC-reID, Occluded-DukeMTMC, Market-1501, MSMT17 and Occluded-REID) verifies the effectiveness of our proposition. The source code and supporting material is available athttps://github.com/zmyme/HDSA-reID.

Finger tapping test is crucial for diagnosing Parkinson's Disease (PD), but manual visual evaluations can result in score discrepancy due to clinicians' subjectivity. Moreover, applying wearable sensors requires making physical contact and may hinder PD patient's raw movement patterns. Accordingly, a novel computer-vision approach is proposed using depth camera and spatial-temporal 3D hand pose estimation to capture and evaluate PD patients' 3D hand movement. Within this approach, a temporal encoding module is leveraged to extend A2J's deep learning framework to counter the pose jittering problem, and a pose refinement process is utilized to alleviate dependency on massive data. Additionally, the first vision-based 3D PD hand dataset of 112 hand samples from 48 PD patients and 11 control subjects is constructed, fully annotated by qualified physicians under clinical settings. Testing on this real-world data, this new model achieves 81.2% classification accuracy, even surpassing that of individual clinicians in comparison, fully demonstrating this proposition's effectiveness. The demo video can be accessed at https://github.com/ZhilinGuo/ST-A2J.

In object counting, objects often exhibit different sizes at different scales, even if they have similar physical sizes in reality. This is particularly true when targeting crowd counting and vehicle counting in intelligent transportation. Failing to model such variations leads to the mismatch between the object size and image scale. To address this problem, existing methods often extract multi-scale features, but they either still generate the single-scale prediction or lack an explicit suppression mechanism to eliminate predictions engendered by inappropriate scales. Our scale analysis manifests that, the single-scale estimation only works well for objects of certain sizes, and a suppression operator is required to isolate the estimation of a specific scale. In this work, we propose a scale-aware counting network termed NSSNet. NSSNet has two key features: it not only i) generates multi-scale predictions but also ii) applies a novel non-scale suppression (NSS) operator to suppress scale-mismatched estimations. NSS is inspired by the widely-used non-maximum suppression (NMS). In contrast to NMS that only reserves the maximum response, NSS filters out those clearly wrong predictions (the remaining predictions may still be from multiple scales). We evaluate NSSNet on four standard crowd and vehicle counting benchmarks and report state-of-the-art performance. We also show the scale adaptability of NSSNet through a controlled multi-scale experiment. Code and pretrained models are available athttps://git.io/nssnet.

Dramatic imaging viewpoint variation is the critical challenge toward action recognition for depth video. To address this, one feasible way is to enhance view-tolerance of visual feature, while still maintaining strong discriminative capacity. Multi-view dynamic image (MVDI) is the most recently proposed 3-D action representation manner that is able to compactly encode human motion information and 3-D visual clue well. However, it is still view-sensitive. To leverage its performance, a discriminative MVDI fusion method is proposed by us via multi-instance learning (MIL). Specifically, the dynamic images (DIs) from different observation viewpoints are regarded as the instances for 3-D action characterization. After being encoded using Fisher vector (FV), they are then aggregated by sum-pooling to yield the representative 3-D action signature. Our insight is that viewpoint aggregation helps to enhance view-tolerance. And, FV can map the raw DI feature to the higher dimensional feature space to promote the discriminative power. Meanwhile, a discriminative viewpoint instance discovery method is also proposed to discard the viewpoint instances unfavorable for action characterization. The wide-range experiments on five data sets demonstrate that our proposition can significantly enhance the performance of cross-view 3-D action recognition. And, it is also applicable to cross-view 3-D object recognition. The source code is available at https://github.com/3huo/ActionView.

We show that explicit modeling of composition rules benefits image cropping. Image cropping is considered a promising way to automate aesthetic composition in professional photography. Existing efforts, however, only model such professional knowledge implicitly, e.g., by ranking from comparative candidates. Inspired by the observation that natural composition traits always follow a specific rule, we propose to learn such rules in a discriminative manner, and more importantly, to incorporate learned composition clues explicitly in the model. To this end, we introduce the concept of the key composition map (KCM) to encode the composition rules. The KCM can reveal the common laws hidden behind different composition rules and can inform the cropping model of what is important in composition. With the KCM, we present a novel cropping-by-composition paradigm and instantiate a network to implement composition-aware image cropping. Extensive experiments on two benchmarks justify that our approach enables effective, interpretable, and fast image cropping.

We show that relation modeling between visual elements matters in cropping view recommendation. Cropping view recommendation addresses the problem of image recomposition conditioned on the composition quality and the ranking of views (cropped sub-regions). This task is challenging because the visual difference is subtle when a visual element is reserved or removed. Existing methods represent visual elements by extracting region-based convolutional features inside and outside the cropping view boundaries, without probing a fundamental question: why some visual elements are of interest or of discard? In this work, we observe that the relation between different visual elements significantly affects their relative positions to the desired cropping view, and such relation can be characterized by the attraction inside/outside the cropping view boundaries and the repulsion across the boundaries. By instantiating a transformer-based solution that represents visual elements as visual words and that models the dependencies between visual words, we report not only state-of-the-art performance on public benchmarks, but also interesting visualizations that depict the attraction and repulsion between visual elements, which may shed light on what makes for effective cropping view recommendation.

Image cropping is a common way to enhance the aesthetic quality of images. Huge industrial demand and the tediousness of image cropping make automatic image cropping a prosperous task. Existing works, however, face two difficulties: objects are easily truncated and key components of images are discarded by the model. The key to solving this problem is to understand the relations between different components of an image. These relations break the limit of spatial distance and reflect the contextual information in images, which help the model decide whether to retain a component. Motivated by this, a patch-related graph module is proposed to model the relations between different patches of an image. The patch-related features are extracted by a graph convolution layer and then fused with the original local features by a proposed gated unit. Moreover, a gradient layer is designed to embed the edge information in the input. The edge-prior input helps the model read the contents of images and reserve the main objects completely. Experimental results show that our model grasps the inter-patch relations well and performs competitively with other state-of-the-art approaches.

Image cropping aims to enhance the aesthetic quality of a given image by removing unwanted areas. Existing image cropping methods can be divided into two groups: candidate-based and candidate-free methods. For candidate-based methods, dense predefined candidate boxes can indeed cover good boxes, but most candidates with low aesthetic quality may disturb the following judgment and lead to an undesirable result. For candidate-free methods, the cropping box is directly acquired according to certain prior knowledge. However, the effect of only one box is not stable enough due to the subjectivity of image cropping. In order to combine the advantages of the above methods and overcome these shortcomings, we need fewer but more representative candidate boxes. To this end, we propose FCRNet, a fully convolutional regression network, which predicts several context-aware cropping boxes in an ensemble manner as candidates. A multi-task loss is employed to supervise the generation of candidates. Unlike previous candidate-based works, FCRNet outputs a small number of context-aware candidates without any predefined box and the final result is selected from these candidates by an aesthetic evaluation network or even manual selection. Extensive experiments show the superiority of our context-aware candidates based method over the state-of-the-art approaches.

Portrait bokeh rendering has become a hot topic in computer vision and graphics in recent years. Existing methods usually suffer from noticeable artifacts around foreground boundaries and unrealistic rendering effects. To tackle these problems, we design a brand new bokeh system in this paper. The system is comprised of three modules: depth estimation, portrait matting, and bokeh rendering. The introduction of the portrait matting module makes it possible to preserve the details of portraits in final rendering results. In bokeh rendering modules, we propose two pixelwise rendering methods which are based on light gathering and light scattering to render realistic bokeh effect. For flexibility and interactivity. We provide two parameter interfaces, i.e., aperture size and bokeh salience to adjust rendering details according to the preferences of different users. Finally, experimental results on our synthetic dataset and real images demonstrate the effectiveness of our proposed method.

One of appealing approaches to counting dense objects, such as crowd, is density map estimation. Density maps, however, present ambiguous appearance cues in congested scenes, rendering infeasibility in identifying individuals and difficulties in diagnosing errors. Inspired by an observation that counting can be interpreted as a two-stage process, i.e., identifying possible object regions and counting exact object numbers, we introduce a probabilistic intermediate representation termed the probability map that depicts the probability of each pixel being an object. This representation allows us to decouple counting into probability map regression (PMR) and count map regression (CMR). We therefore propose a novel decoupled two-stage counting (D2C) framework that sequentially regresses the probability map and learns a counter conditioned on the probability map. Given the probability map and the count map, a peak point detection algorithm is derived to localize each object with a point under the guidance of local counts. An advantage of D2C is that the counter can be learned reliably with additional synthesized probability maps. This addresses important data deficiency and sample imbalanced problems in counting. Our framework also enables easy diagnoses and analyses of error patterns. For instance, we find that, the counter per se is sufficiently accurate, while the bottleneck appears to be PMR. We further instantiate a network D2CNet in our framework and report state-of-the-art counting and localization performance across 6 crowd counting benchmarks. Since the probability map is a representation independent of visual appearance, D2CNet also exhibits remarkable cross-dataset transferability. Code and pretrained models are made available at: https://git.io/d2cnet.

The field of vision-based human hand three-dimensional (3D) shape and pose estimation has attracted significant attention recently owing to its key role in various applications, such as natural humancomputer interactions. With the availability of large-scale annotated hand datasets and the rapid developments of deep neural networks (DNNs), numerous DNN-based data-driven methods have been proposed for accurate and rapid hand shape and pose estimation. Nonetheless, the existence of complicated hand articulation, depth and scale ambiguities, occlusions, and finger similarity remain challenging. In this study, we present a comprehensive survey of state-of-the-art 3D hand shape and pose estimation approaches using RGB-D cameras. Related RGB-D cameras, hand datasets, and a performance analysis are also discussed to provide a holistic view of recent achievements. We also discuss the research potential of this rapidly growing field.

Towards 3D object tracking in point clouds, a novel point-to-box network termed P2B is proposed in an end-to-end learning manner. Our main idea is to first localize potential target centers in 3D search area embedded with target information. Then point-driven 3D target proposal and verification are executed jointly. In this way, the time-consuming 3D exhaustive search can be avoided. Specifically, we first sample seeds from the point clouds in template and search area respectively. Then, we execute permutation-invariant feature augmentation to embed target clues from template into search area seeds and represent them with target-specific features. Consequently, the augmented search area seeds regress the potential target centers via Hough voting. The centers are further strengthened with seed-wise targetness scores. Finally, each center clusters its neighbors to leverage the ensemble power for joint 3D target proposal and verification. We apply PointNet++ as our backbone and experiments on KITTI tracking dataset demonstrate P2B's superiority (~10%'s improvement over state-of-the-art). Note that P2B can run with 40FPS on a single NVIDIA 1080Ti GPU. Our code and model are available at https://github.com/HaozheQi/P2B.

For depth-based 3D action recognition, one essential issue is to represent 3D motion pattern effectively and efficiently. To this end, 3D dynamic voxel (3DV) is proposed as a novel 3D motion representation manner. With 3D space voxelization, the key idea of 3DV is to encode the 3D motion information within depth video into a regular voxel set (i.e., 3DV) compactly, via temporal rank pooling. Each available 3DV voxel intrinsically involves 3D spatial and motion feature for 3D action description. 3DV is then abstracted as a point set and input into PointNet++ for 3D action recognition, in the end-to-end learning way. The intuition for transferring 3DV into the point set form is that, PointNet++ is lightweight and effective for deep feature learning towards point set. Since 3DV may loose appearance clue, a multi-stream 3D action recognition manner is also proposed to learn motion and appearance feature jointly. To extract richer temporal order information of actions, we also split the depth video into temporal segments and encode this procedure in 3DV integrally. The extensive experiments on the well-established benchmark datasets (e.g., NTU RGB+D 120 and NTU RGB+D 60) demonstrate the superiority of our proposition. Impressively, we acquire the accuracy of 82.4% and 93.5% on NTU RGB+D 120 with the cross-subject and cross-setup test setting respectively. 3DV's code is available at https://github.com/3huo/3DV-Action.

Single image depth prediction is a challenging task due to its ill-posed nature and challenges with capturing ground truth for supervision. Large-scale disparity data generated from stereo photos and 3D videos is a promising source of supervision, however, such disparity data can only approximate the inverse ground truth depth up to an affine transformation. To more effectively learn from such pseudo-depth data, we propose to use a simple pair-wise ranking loss with a novel sampling strategy. Instead of randomly sampling point pairs, we guide the sampling to better characterize structure of important regions based on the low-level edge maps and high-level object instance masks. We show that the pair-wise ranking loss, combined with our structure-guided sampling strategies, can significantly improve the quality of depth map prediction. In addition, we introduce a new relative depth dataset of about 21K diverse high-resolution web stereo photos to enhance the generalization ability of our model. In experiments, we conduct cross-dataset evaluation on six benchmark datasets and show that our method consistently improves over the baselines, leading to superior quantitative and qualitative results.

Existing deep siamese trackers are typically built on off-the-shelf CNN models for feature learning, with the demand for huge power consumption and memory storage. This limits current deep siamese trackers to be carried on resource-constrained devices like mobile phones, given factor that such a deployment normally requires cost-effective considerations. In this work, we address this issue by presenting a novel Distilled Learning Framework(DLF) for siamese tracking, which aims at learning tracking model with efficiency and high accuracy. Specifically, we propose two simple yet effective knowledge distillation strategies, denote as point-wise distillation and pairwise distillation, which are designed for transferring knowledge from a more discriminative teacher tracker into a compact student tracker. In this way, cost-effective and high performance tracking could be achieved. Extensive experiments on several tracking benchmarks demonstrate the effectiveness of our proposed method.

Most existing algorithms for depth estimation from single monocular images need large quantities of metric ground-truth depths for supervised learning. We show that relative depth can be an informative cue for metric depth estimation and can be easily obtained from vast stereo videos. Acquiring metric depths from stereo videos are sometimes impracticable due to the absence of camera parameters. In this paper, we propose to improve the performance of metric depth estimation with relative depths collected from stereo movie videos using existing stereo matching algorithm. We introduce a new “relative depth in stereo” (RDIS) dataset densely labeled with relative depths. We first pretrain a ResNet model on our RDIS dataset. Then, we finetune the model on RGB-D datasets with metric ground-truth depths. During our finetuning, we formulate depth estimation as a classification task. This re-formulation scheme enables us to obtain the confidence of a depth prediction in the form of probability distribution. With this confidence, we propose an information gain loss to make use of the predictions that are close to ground-truth. We evaluate our approach on both indoor and outdoor benchmark RGB-D datasets and achieve the state-of-the-art performance.

In this paper, we introduce the idea of blockwise classification to count objects. The current mainstream method for counting objects is to regress the density map or to regress the redundant count map via a deep convolutional neural network (CNN). However, these methods suffer from two critical issues: inaccurately generated regression targets and serious sample imbalances. First, the ground truth density map is generated by convolving the dot map using a Gaussian kernel. Because an inappropriate kernel can cover the background or uncover objects, this approach introduces a form of noise, and therefore results in ambiguities when training the networks. Second, inhomogeneously distributed objects often exist in images, which gives rise to a data collection bias. This leads to a long-tailed distribution of region counts, which is a typical characteristic that occurs with imbalanced samples; therefore, underestimations in high-density regions and overestimations in low-density regions are common. In this paper, we address these two issues within one framework-blockwise count level classification. The intuition behind this idea is that while it may not be possible to provide an exact count of pixels or patches, it is possible to provide a count of a region that falls within a certain interval with high confidence. Our method classifies the count levels of each block produced by nonlinearly quantizing the continuous counts, thus transforming the imbalance of sample patch counts into a class imbalance of count levels. Consequently, an information-entropy-inspired loss can be applied to alleviate this issue. Through ablative studies, we analyze the impact of imbalanced data, Gaussian kernel sizes, quantization errors, and the effectiveness of each module in our method. Without bells and whistles, our method outperforms or performs competitively with other state-of-the-art approaches on seven object-counting benchmarks, including four crowd-counting datasets from ShanghaiTech, WorldExpo'10, UCF-QNRF and UCF_CC_50, one vehicle-counting dataset (TRANCOS), one maize-tassel-counting dataset (MTC), and one challenging sonar fish-counting dataset that we constructed. The results suggest that our framework provides a strong and improved baseline for object counting.

Effective and real-time eyeblink detection is of wide-range applications, such as deception detection, drive fatigue detection, face anti-spoofing. Despite previous efforts, most of existing focus on addressing the eyeblink detection problem under constrained indoor conditions with relative consistent subject and environment setup. Nevertheless, towards practical applications, eyeblink detection in the wild is highly preferred, and of greater challenges. In this paper, we shed the light to this research topic. A labelled eyeblink in the wild dataset (i.e., HUST-LEBW) of 673 eyeblink video samples (i.e., 381 positives, and 292 negatives) is first established. These samples are captured from the unconstrained movies, with the dramatic variation on face attribute, head pose, illumination condition, imaging configuration, etc. Then, we formulate eyeblink detection task as a binary spatial-temporal pattern recognition problem. After locating and tracking human eyes using SeetaFace engine and KCF (Kernelized Correlation Filters) tracker respectively, a modified LSTM model able to capture the multi-scale temporal information is proposed to verify eyeblink. A feature extraction approach that reveals the appearance and motion characteristics simultaneously is also proposed. The experiments on HUST-LEBW reveal the superiority and efficiency of our approach. The comparisons with the existing state-of-the-art methods validate the advantages of our manner for eyeblink detection in the wild.

Image feature correspondence selection is pivotal to many computer vision tasks from object recognition to 3D reconstruction. Although many correspondence selection algorithms have been developed in the past decade, there still lacks an in-depth evaluation and comparison in the open literature, which makes it difficult to choose the appropriate algorithm for a specific application. This paper attempts to fill this gap by evaluating eight competing correspondence selection algorithms including both classical methods and current state-of-the-art ones. In addition to preselected correspondences, we have compared different combinations of detector and descriptor on four standard datasets. The diversity of those datasets cover a wide range of uncertainty factors including zoom, rotation, blur, viewpoint change, JPEG compression, light change, different rendering styles and multiple structures. We have measured the quality of competing correspondence selection algorithms in terms of four performance metrics -i.e., precision, recall, F-measure and efficiency. Moreover, we propose to combine the strengths of eight competing methods by combining their correspondence selection results. Extensive experimental results are reported to demonstrate the superiority of several fusion strategies to individual methods, which suggests the possibility of adaptively combining those methods for even better performance.

In this paper, we mainly investigate two issues for sequence labeling, namely, label imbalance and noisy data that are commonly seen in the scenario of named entity recognition (NER) and are largely ignored in the existing works. To address these two issues, a new method termed robust sequence labeling (RoSeq) is proposed. Specifically, to handle the label imbalance issue, we first incorporate label statistics in a novel conditional random field (CRF) loss. In addition, we design an additional loss to reduce the weights of overwhelming easy tokens for augmenting the CRF loss. To address the noisy training data, we adopt an adversarial training strategy to improve model generalization. In experiments, the proposed RoSeq achieves the state-of-the-art performances on CoNLL and English Twitter NER-88.07% on CoNLL-2002 Dutch, 87.33% on CoNLL-2002 Spanish, 52.94% on WNUT-2016 Twitter, and 43.03% on WNUT-2017 Twitter without using the additional data.

Feature correspondence selection is pivotal to many feature-matching based tasks in computer vision. Searching spatially k-nearest neighbors is a common strategy for extracting local information in many previous works. However, there is no guarantee that the spatially k-nearest neighbors of correspondences are consistent because the spatial distribution of false correspondences is often irregular. To address this issue, we present a compatibility-specific mining method to search for consistent neighbors. Moreover, in order to extract and aggregate more reliable features from neighbors, we propose a hierarchical network named NM-Net with a series of graph convolutions that is insensitive to the order of correspondences. Our experimental results have shown the proposed method achieves the state-of-the-art performance on four datasets with various inlier ratios and varying numbers of feature consistencies.

Visual counting, a task that predicts the number of objects from an image/video, is an open-set problem by nature, i.e., the number of population can vary in [0,+∞) in theory. However, the collected images and labeled count values are limited in reality, which means only a small closed set is observed. Existing methods typically model this task in a regression manner, while they are likely to suffer from an unseen scene with counts out of the scope of the closed set. In fact, counting is decomposable. A dense region can always be divided until the count values of sub-regions are within the previously observed closed set. Inspired by this idea, we propose a simple but effective approach, Spatial Divide-and-Conquer Network (S-DCNet). S-DCNet learns to classify closed-set counts and can generalize to open-set counts via S-DC. S-DCNet is also efficient. To avoid repeatedly computing sub-region convolutional features, S-DC is executed on the feature map instead of on the input image. S-DCNet achieves the state-of-the-art performance on three crowd counting datasets (ShanghaiTech, UCF_CC_50 and UCF-QNRF), a vehicle counting dataset (TRANCOS) and a plant counting dataset (MTC). Compared to the previous best methods, S-DCNet brings a 20.2% relative improvement on the ShanghaiTechPart B, 20.9% on the UCF-QNRF, 22.5% on the TRANCOS and 15.1% on the MTC. Code has been made available at: https://github.com/xhp-hust-2018-2011/S-DCNet.

For 3D hand and body pose estimation task in depth image, a novel anchor-based approach termed Anchor-to-Joint regression network (A2J) with the end-to-end learning ability is proposed. Within A2J, anchor points able to capture global-local spatial context information are densely set on depth image as local regressors for the joints. They contribute to predict the positions of the joints in ensemble way to enhance generalization ability. The proposed 3D articulated pose estimation paradigm is different from the state-of-the-art encoder-decoder based FCN, 3D CNN and point-set based manners. To discover informative anchor points towards certain joint, anchor proposal procedure is also proposed for A2J. Meanwhile 2D CNN (i.e., ResNet- 50) is used as backbone network to drive A2J, without using time-consuming 3D convolutional or deconvolutional layers. The experiments on 3 hand datasets and 2 body datasets verify A2J's superiority. Meanwhile, A2J is of high running speed around 100 FPS on single NVIDIA 1080Ti GPU.

In this paper, we study the problem of self-supervised monocular depth prediction. Owing to the fact that a vast quantity of expensive ground-truth depth data is required in supervised deep learning methods, self-supervised deep learning methods are what we apply to and are more approachable as well. Specifically, we base our model on a monocular disparity network that generates disparity images by training with an image reconstruction loss. Because binocular images implicitly provide epipolar geometry constraints, we find that binocular depth estimators always perform better than monocular ones. Therefore, we propose a novel module (i.e., BinoBoost) to boost our monocular disparity network with binocular guidance during training. In particular, the binocular disparity network is trained in a similar way to the base model to generate proxy ground truth disparity. We drive the outputs of the base model to be the same as relatively better outputs produced by the binocular disparity network. By training our base model and BinoBoost in an end-to-end fashion, we improve the performance on our base model and achieve the state-of-the-art results on the KITTI dataset.

The aggregation of hierarchical information is vital for saliency detection. To achieve this, most existing saliency detectors apply various network structures to fuse features. But most of them utilize shallow skip connections and only concentrate on the final results, which can not guarantee the model to learn the rich and accurate contextual information. To address these problems, we propose a network with deep layer aggregation and multi-layer intermediate supervision. We utilize deep layer aggregation to fuse features iteratively and hierarchically across layers to obtain richer information. Then we add multi-layer intermediate supervision on each side-output layer to capture more accurate contextual information. We evaluate our method on six benchmark datasets under various metrics and it achieves the new state-of-the-art.

Monocular depth estimation is a widely studied computer vision problem with a vast variety of applications. In this paper, we formulate it as a pixel-wise classification task and use a mean-variance loss for robust depth estimation via distribution learning. More precisely, the mean-variance loss is composed of a mean loss that penalizes the difference between the mean of predicted depth distribution and the ground-truth depth, and a variance loss that penalizes the variance of predicted depth distribution to obtain a more focused distribution. The mean-variance loss is jointly trained with the soft-max loss to supervise a Deep Convolutional Neural Networks (DCNN) for depth estimation. Experimental results on the NYUDv2 dataset show that the proposed method outperforms previous state-of-the-art approaches.

Toward the medical and living healthcare for the elderly and patients, fall from bed is a critical accident that may lead to serious injuries. To alleviate this, an essential problem is to detect this event in time for earning the rescue time. Although some efforts that resort to the wearable devices and smart healthcare room have already been paid to address this problem, the performance is still not satisfactory enough for the practical applications. In this paper, a novel fall from a bed detection method is proposed. In particular, the depth camera is used as the visual sensor due to its insensitivity to illumination variation and capacity of privacy protection. To characterize the human activity well, an effective human upper body detection approach able to extract human head and upper body center is proposed using random forest. Compared with the existing widely used human body parsing methods (e.g., Microsoft Kinect SDK or OpenNI SDK), our proposition can still work reliably when human-bed interaction happens. According to the motion information of human upper body, the fall from bed detection task is formulated as a two-class classification problem. Then, it is solved using the large margin nearest neighbor classification approach. Our method can meet the real-time running requirement with the normal computer. In experiments, we construct a fall from bed detection data set that contains the samples from 42 volunteers (26 males and 16 females) for test. The experimental results demonstrate the effectiveness and efficiency of our proposition.

Aligning 2.5D views has been extensively explored in the past decades, where most prior works have concentrated on object data with complex structures. This paper presents a method to align real-word scene scans with challenging features such as noise, poor geometric information, and highly repeatable patterns. Our method consists of two modules: pairwise and multiview alignments. Key to the proposed pairwise alignment method is the rotational contour signature geometric feature and voting-based correspondence selection algorithm. The former promises strong discriminative power for 2.5D scene data, while the latter affords high-quality correspondences via a voting process for all raw feature matches using L2distance and point pair affinity constraints. For the multiview alignment method, we first use a connected graph algorithm to establish the connections of all 2.5D views for coarse merging; then, we propose a shape-growing iterative closest point algorithm for further refinement. Experiments are conducted on scene point cloud datasets addressing both the indoor and outdoor scenarios, whereby we demonstrate that the proposed pairwise alignment method clearly outperforms the state of the art. Moreover, the proposed multiview alignment method manages to put multiple unordered 2.5D scene fragments into a unified coordinate system automatically, accurately, and efficiently.

The whole-sky imager has been increasingly used for ground-based cloud automatic observation. Many approaches based on image processing have been applied to detect or classify clouds in whole-sky images (WSIs). However, most of the studies only focus on image segmentation for cloud detection or image classification for cloud recognition separately. The cloud detection only does the binary segmentation (sky and cloud) without cloud types, while the cloud recognition only gives the single image-level label without cloud coverage. In this paper, a fine-grained cloud detection and recognition task with a solution is proposed to fill the gap, which can simultaneously detect and classify clouds in a WSI. It can be regarded as a pixel-level fine-grained dense prediction for images. First, a new data set is built with pixel-level annotation of nine different types. Then, a solution based on supervised learning is proposed, in which the pixel-level prediction problem is converted to a superpixel classification problem. Multiview features are extracted, including color, inside texture, neighbor texture, and global relation, to represent the superpixels. Moreover, a class-specific feature space transformation method based on metric learning and subspace alignment is proposed to overcome the challenge brought by the high similarity among cloud types and the feature shifting. Finally, several experiments have verified that our approach is effective to the challenging new task and also outperforms some other methods in the normal tasks of cloud detection and cloud classification, respectively.

In this paper, we study a new problem in the scenario of sequences labeling. To be exact, we consider that the training data are with annotation of various degrees, namely, fully labeled, unlabeled, and partially labeled sequences. The learning with fully un/labeled sequence refers to the standard setting in traditional un/supervised learning, and the proposed partially labeling specifies the subject that the element does not belong to. The partially labeled data are cheaper to obtain compared with the fully labeled data though it is less informative, especially when the tasks require a lot of domain knowledge. To solve such a practical challenge, we propose a novel deep conditional random field (CRF) model which utilizes an end-to-end learning manner to smoothly handle fully/un/partially labeled sequences within a unified framework. To the best of our knowledge, this could be one of the first works to utilize the partially labeled instance for sequence labeling, and the proposed algorithm unifies the deep learning and CRF in an end-to-end framework. Extensive experiments show that our method achieves state-of-the-art performance in two sequence labeling tasks on some popular data sets.

In this paper we study the problem of monocular relative depth perception in the wild. We introduce a simple yet effective method to automatically generate dense relative depth annotations from web stereo images, and propose a new dataset that consists of diverse images as well as corresponding dense relative depth maps. Further, an improved ranking loss is introduced to deal with imbalanced ordinal relations, enforcing the network to focus on a set of hard pairs. Experimental results demonstrate that our proposed approach not only achieves state-of-the-art accuracy of relative depth perception in the wild, but also benefits other dense per-pixel prediction tasks, e.g., metric depth estimation and semantic segmentation.

The goal of this paper is to estimate the population of fishes in sonar images. Compared to natural images, sonar images present substantially different visual characteristics. Fishes in sonar images exhibit unreliable appearance cues, expose under imaging noise, vary significantly in shape and size. These pose great challenges for counting even for a human expert. In Computer Vision, a possible solution to this task is object counting with deep networks. This paradigm is typically formulated as a regression problem. The regression, however, greatly suffers from the issue of sample imbalance caused by fish variations in size and density, leading to underestimates in high-density regions and over-estimates in low-density regions. To address this, we build upon a recent local counts regression network and propose two novel losses to regularize a modified l1 loss with slack constraints. In particular, a challenging sonar fish counting dataset with 537 images and manually labeled dotted annotations is constructed. Experimental results on the dataset justify the effectiveness of our proposition and show improved performance of our method over other state-of-the-art approaches.

Correspondence selection aiming at seeking correct relationships between two images is a fundamental and critical task in computer vision. This paper attempts to select consistent correspondences in the context of dynamic scenarios where multiple matching consistencies are normally incorporated. To this end, we present a grid-based game-theoretic matching (Grid-GTM) method which is divided into three processes, i.e., grid matching, local games and enrichment. Specifically, grid matching translates the multi-consistency problem into several independent single-consistency problems to decrease difficulties of selection and boost the efficiency. Local games extended under the guidance of a novel payoff function guarantee that mismatches are effectively removed. Enrichment is added to recover correct matches neglected by local games. Crucially, our approach achieves the state-of-the-art performance compared with seven algorithms in comprehensive evaluations. In addition, we construct a dataset that involves multiple consistencies under three different scenes in this paper.

Crop cultivar identification is an important aspect in agricultural systems. Traditional solutions involve excessive human interventions, which is labor-intensive and timeconsuming. In addition, cultivar identification is a typical task of fine-grained visual categorization (FGVC). Compared with other common topics in FGVC, studies of this problem are somewhat lagging and limited. In this paper, targeting four Chinese maize cultivars of Jundan No.20, Wuyue No.3, Nongda No.108, and Zhengdan No.958, we first consider the problem of identifying the maize cultivar based on its tassel characteristics by computer vision. In particular, a novel fine-grained maize cultivar identification data set termed HUST-FG-MCI that contains 5000 images is first constructed. To better capture the textual differences in a weakly supervised manner, we proposed an effective deep convolutional neural network and Fisher vector (FV)based feature encoding mechanism. The mechanism tends to highlight subtle object patterns via filter-specific convolutional representations and thus provides strong discrimination for cultivar identification. Experimental results demonstrate that our method outperforms other state-of-the-art approaches. We show also that FV encoding can weaken the linear dependency between convolutional activations, redundant filters exist in the convolutional layer, and high accuracy can be maintained with relatively low-dimensional convolutional features and one or two Gaussian components in FV.

We show that it is possible to achieve high-quality domain adaptation without explicit adaptation. The nature of the classification problem means that when samples from the same class in different domains are sufficiently close, and samples from differing classes are separated by large enough margins, there is a high probability that each will be classified correctly. Inspired by this, we propose an embarrassingly simple yet effective approach to domain adaptation-only the class mean is used to learn class-specific linear projections. Learning these projections is naturally cast into a linear-discriminant-analysis-like framework, which gives an efficient, closed form solution. Furthermore, to enable to application of this approach to unsupervised learning, an iterative validation strategy is developed to infer target labels. Extensive experiments on cross-domain visual recognition demonstrate that, even with the simplest formulation, our approach outperforms existing non-deep adaptation methods and exhibits classification performance comparable with that of modern deep adaptation methods. An analysis of potential issues effecting the practical application of the method is also described, including robustness, convergence, and the impact of small sample sizes.

The local reference frame (LRF), as an independent coordinate system constructed on the local 3D surface, is broadly employed in 3D local feature descriptors. The benefits of the LRF include rotational invariance and full 3D spatial information, thereby greatly boosting the distinctiveness of a 3D feature descriptor. There are numerous LRF methods in the literature; however, no comprehensive study comparing their repeatability and robustness performance under different application scenarios and nuisances has been conducted. This paper evaluates eight state-of-the-art LRF proposals on six benchmarks with different data modalities (e.g., LiDAR, Kinect, and Space Time) and application contexts (e.g., shape retrieval, 3D registration, and 3D object recognition). In addition, the robustness of each LRF to a variety of nuisances, including varying support radii, Gaussian noise, outliers (shot noise), mesh resolution variation, distance to boundary, keypoint localization error, clutter, occlusion, and partial overlap, is assessed. The experimental study also measures the performance under different keypoint detectors, descriptor matching performance when using different LRFs and feature representation combinations, as well as computational efficiency. Considering the evaluation outcomes, we summarize the traits, advantages, and current limitations of the tested LRF methods.

This paper presents a thorough evaluation of several widely-used 3D correspondence grouping algorithms, motived by their significance in vision tasks relying on correct feature correspondences. A good correspondence grouping algorithm is desired to retrieve as many as inliers from initial feature matches, giving a rise in both precision and recall. Towards this rule, we deploy the experiments on three benchmarks respectively addressing shape retrieval, 3D object recognition and point cloud registration scenarios. The variety in application context brings a rich category of nuisances including noise, varying point densities, clutter, occlusion and partial overlaps. It also results to different ratios of inliers and correspondence distributions for comprehensive evaluation. Based on the quantitative outcomes, we give a summarization of the merits/demerits of the evaluated algorithms from both performance and efficiency perspectives.

Domain adaption (DA) allows machine learning methods trained on data sampled from one distribution to be applied to data sampled from another. It is thus of great practical importance to the application of such methods. Despite the fact that tensor representations are widely used in Computer Vision to capture multi-linear relationships that affect the data, most existing DA methods are applicable to vectors only. This renders them incapable of reflecting and preserving important structure in many problems. We thus propose here a learning-based method to adapt the source and target tensor representations directly, without vectorization. In particular, a set of alignment matrices is introduced to align the tensor representations from both domains into the invariant tensor subspace. These alignment matrices and the tensor subspace are modeled as a joint optimization problem and can be learned adaptively from the data using the proposed alternative minimization scheme. Extensive experiments show that our approach is capable of preserving the discriminative power of the source domain, of resisting the effects of label noise, and works effectively for small sample sizes, and even one-shot DA. We show that our method outperforms the state-of-the-art on the task of cross-domain visual recognition in both efficacy and efficiency, and particularly that it outperforms all comparators when applied to DA of the convolutional activations of deep convolutional networks.

Accurate ground-based cloud image categorization is a critical but challenging task that has not been well addressed. One of the essential issues that affect the performance is to extract the representative visual features. Nearly all of the existing methods rely on the hand-crafted descriptors (e.g., local binary patterns, CENsus TRsansform hISTogram, and scale-invariant feature transform). Their limited discriminative power indeed leads to the unsatisfactory performance. To alleviate this, we propose “DeepCloud” as a novel cloud image feature extraction approach by resorting to the deep convolutional visual features. In the recent years, the deep convolutional neural network (CNN) has achieved the promising results in lots of computer vision and image understanding fields. Nevertheless, it has not been applied to cloud image classification yet. Thus, we actually pay the first effort to fill this blank. Since cloud image classification can be attributed to a multi-instance learning problem, simply employing the convolutional features within CNN cannot achieve the promising result. To address this, Fisher vector encoding is applied to executing the spatial feature aggregation and high-dimensional feature mapping on the raw deep convolutional features. Moreover, the hierarchical convolutional layers are used simultaneously to capture the fine textural characteristics and high-level semantic information in the unified manner. To further leverage the performance, a cloud pattern mining and selection method are also proposed. It targets at finding the discriminative local patterns to better distinguish the different kinds of clouds. The experiments on a challenging ground-based cloud image data set demonstrate the superiority of the proposition over the state-of-the-art methods.

Cultivar identification is an important aspect in agriculture and also a typical task of fine-grained visual categorization (FGVC). In comparison with other common topics in FGVC, studies on this problem are somewhat lagged and limited. In this paper, targeting four Chinese maize cultivars of Jundan No.20, Wuyue No.3, Nongda No.108, and Zhengdan No.958, we first consider the problem of identifying the maize cultivar based on its tassel characteristics. Technically, an effective convolutional neural network (CNN) based feature encoding pipeline that allows integration of deep CNN based column feature extraction, filter-specific Fisher vector (FV) encoding and mutual information (MI) based filter selection is proposed to better address this problem. In particular, a novel fine-grained maize cultivar identification dataset termed MCI-4000 that contains 4000 images is first constructed by our team. Experimental results demonstrate that our method outperforms other stat-of-the-art approaches by at least 5% in accuracy. We also show that, there exists redundant filters in the last convolutional layer, and high accuracy can be achieved with only relatively low-dimensional column features and a small number of Gaussian components in FV.

This paper presents a novel local surface descriptor called rotational contour signatures (RCS) for 3D rigid objects. RCS comprises several signatures that characterize the 2D contour information derived from 3D-to-2D projection of the local surface. The inspiration of our encoding technique comes from that, viewing towards an object, its contour is an effective and robust cue for representing its shape. In order to achieve a comprehensive geometry encoding, the local surface is continually rotated in a predefined local reference frame (LRF) so that multi-view information is obtained. Experiments on two publicly available datasets demonstrate the effectiveness and robustness of the proposed descriptor. Further, comparisons with five state-of-the-art descriptors show the superiority of our RCS descriptor.

Attributes now play a vital role for characterizing a crowded scene. Compared to low-level visual features, processing informed by attributes can capture rich semantic information. However, to effectively assign attributes to a crowded scene still remains a challenging task. In this letter, inspired by a recently proposed zero-shot learning framework, a novel attribute assignment method that maps low-level features to predefined attributes is proposed. In particular, we propose to exploit the attribute dependency during the phase of attribute assignment, which can be regarded as our main contribution. In addition, to further enhance the performance, an effective low-level feature extraction mechanism is also proposed. More precisely, appearance and motion features are first simultaneously extracted from several sampled video frames and corresponding optical flow fields via deep convolutional neural network and then, respectively, aggregated by using Fisher vector encoding to form the low-level representation of crowded scenes. Experimental results on the challenging WWW dataset demonstrate that both the proposed attribute assignment method and the low-level feature extraction mechanism outperform the state of the art.

Despite the previous efforts of object proposals, the detection rates of the existing approaches are still not satisfactory enough. To address this, we propose Adobe Boxes to efficiently locate the potential objects with fewer proposals, in terms of searching the object adobes that are the salient object parts easy to be perceived. Because of the visual difference between the object and its surroundings, an object adobe obtained from the local region has a high probability to be a part of an object, which is capable of depicting the locative information of the proto-object. Our approach comprises of three main procedures. First, the coarse object proposals are acquired by employing randomly sampled windows. Then, based on local-contrast analysis, the object adobes are identified within the enlarged bounding boxes that correspond to the coarse proposals. The final object proposals are obtained by converging the bounding boxes to tightly surround the object adobes. Meanwhile, our object adobes can also refine the detection rate of most state-of-the-art methods as a refinement approach. The extensive experiments on four challenging datasets (PASCAL VOC2007, VOC2010, VOC2012, and ILSVRC2014) demonstrate that the detection rate of our approach generally outperforms the state-of-the-art methods, especially with relatively small number of proposals. The average time consumed on one image is about 48 ms, which nearly meets the real-time requirement.

Image blurring significantly degrades the image recognition performance. In this paper, we novelly address the blurred image recognition task from the perspective of domain adaptation (DA). The scenario is that, the training set (source domain) only comprises of the labelled clear images, and the test set (target domain) is composed of the unlabelled blurred images. DA is executed to eliminate the domain shift by subspace alignment. In this way, the clear and blurred image domains are pushed closer in the feature space. The supervised LMDR metric learning method is employed by us to construct the source domain subspace for further performance enhancement, compared to the unsupervised one (i.e., PCA). The experimental results on two datasets demonstrate that, the proposed DA-based blurred image recognition mechanism can significantly enhance the performance of different kinds of visual descriptors, especially when the blurring degree is strong.

Ground-based cloud image categorization is an essential and challenging task in automatic sky and cloud observation field. Till now, it still has not been well addressed in both meteorology and image processing communities, due to the large variation of cloud appearance. One feasible way to solve this is to find more discriminative visual representation to characterize the different kinds of clouds. Many efforts have been paid in this way. However, to our knowledge, most of the existing methods only resort to the hand-craft visual descriptors (e.g., LBP, CENTRIST and color histogram). The resulting performance is unfortunately not satisfied enough. Inspired by the great success of deep convolutional neural networks (CNN) in large-scale image classification task (e.g., ImageNet challenge), we first propose to transfer CNN to solve our relative small-scale cloud classification issue. The experiments on two challenging cloud datasets demonstrate that, using the deep convolutional visual features generated by CNN can significantly outperform all the state-of-the-art methods in most cases. Another important contribution of our work is that, we find that applying Fisher Vector (FV) to encoding the off-the-shelf CNN features can further leverage the performance.

Blurred image recognition is still remaining as a challenging task, while with the wide applications. One principal way for solving this problem is to extract the blur-invariant visual descriptor. To this end, local phase quantization (LPQ) was ever proposed, and achieved promising results. In this paper, to further enhance LPQ's performance, we propose to apply Fisher Vector (FV) encoding approach to acquire the mid-level blurred image representation. To our knowledge, it is the first time that the descriptive power of FV for blurred image recognition has been investigated. Instead of being extracted holistically from the whole image as previously, LPQ is acquired in a densely sampled way. That is, a sliding sub-window will screen the image with certain vertical and horizontal strides. LPQs are then extracted from all the resulting sub-windows respectively. In addition, to maintain local spatial structure information, each sub-window will be divided into finer cells. After being FV encoded, the local LPQs are aggregated using sum-pooling to generate the image signature. The experimental results on three datasets demonstrate that FV can enhance LPQ's performance significantly, and our proposition also outperforms the other blur-invariant descriptors by large margins in most cases.

Both image registration and fusion are essential steps to produce high-resolution multispectral images in remote sensing. Traditionally, they are viewed as two independent processes. As a result, the registration errors ignored in the fusion process can significantly affect the fusion quality. In this context, an iterative optimization approach, which jointly considers the registration and fusion processes, is proposed for panchromatic (PAN) and multispectral (MS) images. Given a registration method and a fusion method, the joint optimization process is described as finding the optimal registration parameters to gain the optimal fusion performance. In our approach, the downhill simplex algorithm is adopted to refine the registration parameters iteratively. Experiments on a set of PAN and MS images of ZY-3 and GeoEye-1 show that the proposed approach outperforms several competing ones in terms of registration accuracy and fusion quality.

Detecting and matching image features is a fundamental task in video analytics and computer vision systems. It establishes the correspondences between two images taken at different time instants or from different viewpoints. However, its large computational complexity has been a challenge to most embedded systems. This paper proposes a new FPGA-based embedded system architecture for feature detection and matching. It consists of scale-invariant feature transform (SIFT) feature detection, as well as binary robust independent elementary features (BRIEF) feature description and matching. It is able to establish accurate correspondences between consecutive frames for 720-p (1280x720) video. It optimizes the FPGA architecture for the SIFT feature detection to reduce the utilization of FPGA resources. Moreover, it implements the BRIEF feature description and matching on FPGA. Due to these contributions, the proposed system achieves feature detection and matching at 60 frame/s for 720-p video. Its processing speed can meet and even exceed the demand of most real-life real-time video analytics applications. Extensive experiments have demonstrated its efficiency and effectiveness.

This paper presents a new method for salient object detection based on a sophisticated appearance comparison of multisize superpixels. Those superpixels are modeled by multivariate normal distributions in CIE-Lab color space, which are estimated from the pixels they comprise. This fitting facilitates an efficient application of the Wasserstein distance on the Euclidean norm ( [Formula: see text]) to measure perceptual similarity between elements. Saliency is computed in two ways. On the one hand, we compute global saliency by probabilistically grouping visually similar superpixels into clusters and rate their compactness. On the other hand, we use the same distance measure to determine local center-surround contrasts between superpixels. Then, an innovative locally constrained random walk technique that considers local similarity between elements balances the saliency ratings inside probable objects and background. The results of our experiments show the robustness and efficiency of our approach against 11 recently published state-of-the-art saliency detection methods on five widely used benchmark data sets.

We present a new segment-based method for saliency detection based on multi-size superpixels that combines local and global saliency cues. We extract superpixels at several scales and represent each superpixel with a normal distribution in CIE-Lab space estimated from its associated pixels. Global saliency is computed by grouping similar superpixels to estimate the spatial distribution of colors, while local saliency detection is achieved by determining the center-surround contrast of neighboring superpixels. Both methods rely on the Wasserstein distance on L2norm (W2) to measure perceptual (dis-)similarity between superpixels. Additionally, we propose a Saliency Flow technique to refine the local saliency map. Our approach uses very few empirical parameters and outperforms 6 recent state-of-the-art saliency detection methods in terms of several evaluations on a widely used benchmark.

Presence is a service that allows a user to be informed about the reachability, availability, and willingness of communication of another user. Presence service has become a key enabler for many popular applications such as instant messaging and push-to-talk. Traffic to a presence server is resulted by user behaviors such as login/logout, online status modification, and automatic status refresh by client application. Mathematical models are proposed in this paper to study the user behaviors associated with a presence server during the time of a day such that traffic characteristics to a presence server can be inferred. The correctness of our estimation and analysis are verified by extensive simulations using Matlab. This study on the relationship between user behaviors and traffic to a presence server can help network operators to plan network capacity, optimize server performance and detect traffic anomaly.

In this paper, an entropic thresholding method based on the gray-level spatial correlation (GLSC) histogram defined by ourselves is presented. Compared with traditional two-dimensional histogram, we take into account the image local property in a different way by GLSC histogram. In experiment, we make comparison of the proposed method with two-dimensional entropic thresholding method proposed by Abutaleb and one-dimensional entropic thresholding method proposed by Kapur. The experiment demonstrates that generally our method could yield equivalent or even better result than Abutaleb’s method while saving time remarkably and perform much better than Kapur’s method without too more time consumption.

In this paper, a multi-sensor data fusion method based on dynamic fuzzy neural network (DFNN) for object recognition is proposed.DFNN is composed of two individual fuzzy neural networks. During the practical recognition process, one fuzzy neural network is used for recognition while the other is tracking trained. At the appropriate time the role of the two networks can be exchanged according to certain switching rule. The fusion recognition system is composed of two layers. At the first layer, the features extracted from middle wave and long wave infrared images are fused by DFNN to detect potential regions which may contain objects. And then the features extracted from visible image are utilized to make recognition in these potential regions based on DFNN at the second layer. The experiment demonstrates the efficiency of the proposed method.