The goal of RGB-Thermal (RGB-T) tracking is to utilize the synergistic and complementary strengths of RGB and TIR modalities to enhance tracking in diverse situations, with cross-modal interaction being a crucial element. Earlier methods often simply combine the features of the RGB and TIR search frames, leading to a coarse interaction that also introduced unnecessary background noise. Many other approaches sample candidate boxes from search frames and apply different fusion techniques to individual pairs of RGB and TIR boxes, which confines cross-modal interactions to local areas and results in insufficient context modeling. Additionally, mining video temporal contexts is also under-explored in RGB-T tracking. To alleviate these limitations, we propose a novel Template-Bridged Search region Interaction (TBSI) module that exploits templates as the medium to bridge the cross-modal interaction between RGB and TIR search regions by gathering and distributing target-relevant object and environment contexts. An Illumination Guided Fusion (IGF) module is designed to adaptively fuse RGB and TIR search region tokens with a global illumination factor. Furthermore, in the inference stage, we also propose an efficient Target-Preserved Template Updating (TPTU) strategy, leveraging the temporal context within video sequences to accommodate the target's appearance change. Our proposed modules are integrated into a ViT backbone for joint feature extraction, search-template matching, and cross-modal interaction. Extensive experiments on three popular RGB-T tracking benchmarks demonstrate our method achieves new state-of-the-art performances.

The anatomies in low-dose computer tomography (LDCT) are usually distorted during the zooming-in observation process due to the small amount of quantum. Super-resolution (SR) methods have been proposed to enhance qualities of LDCT images as post-processing approaches without increasing radiation damage to patients, but suffered from incorrect prediction of degradation information and incomplete leverage of internal connections within the 3D CT volume, resulting in the imbalance between noise removal and detail sharpening in the super-resolution results. In this paper, we propose a novel LDCT SR network where the degradation information self-parsed from the LDCT slice and the 3D anatomical information captured from the LDCT volume are integrated to guide the backbone network. The prior degradation estimator (PDE) is proposed following the contrastive learning strategy to estimate the degradation features in the LDCT images without paired low-normal dose CT images. The self-guidance fusion module (SGFM) is designed to capture anatomical features with internal 3D consistencies between the squashed images along the coronal, sagittal, and axial views of the CT volume. Finally, the features representing degradation and anatomical structures are integrated to recover the CT images with higher resolutions. We apply the proposed method to the 2016 NIH-AAPM Mayo Clinic LDCT Grand Challenge dataset and our collected LDCT dataset to evaluate its ability to recover LDCT images. Experimental results illustrate the superiority of our network concerning quantitative metrics and qualitative observations, demonstrating its potential in recovering detail-sharp and noise-free CT images with higher resolutions from the practical LDCT images.

Most of the existing 3D talking face synthesis methods suffer from the lack of detailed facial expressions and realistic head poses, resulting in unsatisfactory experiences for users. In this article, we propose a novel pose-aware 3D talking face synthesis method with a novel geometry-guided audio-vertices attention. To capture more detailed expression, such as the subtle nuances of mouth shape and eye movement, we propose to build hierarchical audio features including a global attribute feature and a series of vertex-wise local latent movement features. Then, in order to fully exploit the topology of facial models, we further propose a novel geometry-guided audio-vertices attention module to predict the displacement of each vertex by using vertex connectivity relations to take full advantage of the corresponding hierarchical audio features. Finally, to accomplish pose-aware animation, we expand the existing database with an additional pose attribute, and a novel pose estimation module is proposed by paying attention to the whole head model. Numerical experiments demonstrate the effectiveness of the proposed method on realistic expression and head movements against state-of-the-art methods.

We present the pioneering Large Visual Motion Model (LVMM), meticulously engineered to analyze the intrinsic dynamics encapsulated within real-world imagery. Our model, fortified with a wealth of prior knowledge extracted from billions of image pairs, demonstrates promising results in predicting a diverse spectrum of scene dynamics. As a result, it can infuse any generic image with authentic dynamic effects, enhancing its visual allure. Project page: https://github.com/densechen/LVMM.

In focusing tasks on moving targets, traditional methods that rely on maximizing contrast struggle to capture moving objects due to insufficient focusing speed. Deep learning-based methods have attempted to directly predict the optimal focal length for the target; however, due to low prediction accuracy, they often lead to out-of-focus situations when capturing moving objects. In recent years, some approaches have utilized reinforcement learning to automatically explore focal length adjustment patterns, thus achieving better results than traditional methods. However, these approaches have not considered the motion characteristics of the targets, leading to a need for further improvement in focusing performance. To overcome these limitations, we introduce a motion-based feature and deep reinforcement learning-driven autofocus algorithm named MF-DRLAF (Motion Features based Deep Reinforcement Learning Autofocus Model) for moving targets. This novel method tracks the object, predicts its motion state through feature extraction, and uses deep reinforcement learning to dynamically adjust the focus. We utilize a self-attention mechanism to adaptively learn various motion patterns and employ a feature pool structure to enhance processing efficiency. Experiments and real-world testing on a Google Pixel3 demonstrate that our approach significantly enhances autofocus performance on moving objects, highlighting its potential for broader imaging applications. This approach offers a promising direction for future development in autofocus technology.

Traditional autofocus methods search for the optimal focal distance (FD) by evaluating image quality from focal stacks, resulting in time-consuming focusing processes. Recently, deep learning has being adopted for single-shot autofocus methods, which can predict the optimal FD directly from a single input image. However, these methods often suffer from low prediction accuracy due to the lack of global features and structured global supervisory information, as they rely solely on the image's region of interest (ROI) as input and a single value for supervision. We propose a deep learning network named MPFS (Multi-Head Network with Per-Pixel Focal Distance Supervision), which takes a full-frame photograph as input and uses the optimal focal distance per pixel for supervision, this method effectively addresses the issues of missing global features and insufficient supervisory information by leveraging these enhancements. Additionally, the network integrates current camera focal distance information to mitigate the scale ambiguity caused by the lack of absolute scale information. To validate the effectiveness of the proposed method, we designed an experiment using a dataset annotated with optimal FD per pixel. Experimental results on this dataset indicate that our approach achieves a 0.22 decrease in the Mean Absolute Error (Mae) metric compared to the state-of-the-art models, with improvements of 0.02 and 0.004 in$\boldsymbol{d}_{\mathbf{1}}$and$\boldsymbol{d}_{\mathbf{2}}$metrics.

Seismic data often suffer from missing traces, and traditional reconstruction methods are cumbersome in parameterization and struggle to handle large-scale missing data. While deep learning has shown powerful reconstruction capabilities, convolutional neural networks’ (CNNs) point-to-point reconstruction may not fully cover the distribution of the entire dataset and may suffer performance degradation under complex missing patterns. In response to this challenge, we propose a novel diffusion model reconstruction framework tailored for 3-D seismic data. To facilitate 3-D seismic data reconstruction using diffusion models, we introduce conditional constraints into the diffusion model, constraining the generated data of the diffusion model based on the input data to be reconstructed. We introduce a 3-D neural network architecture into the diffusion model and refine the diffusion model’s generation process by incorporating existing parts of the data into the generation process, resulting in reconstructions with higher consistency. Through ablation studies determining optimal parameter values, although the sampling time is longer, our method exhibits superior reconstruction accuracy when applied to both field datasets and synthetic datasets, effectively addressing a wide range of complex missing patterns. Our implementation is available athttps://github.com/WAL-l/SeisFusion.

It is well believed that the higher uncertainty in a word of the caption, the more inter-correlated context information is required to determine it. However, current image captioning methods usually consider the generation of all words in a sentence sequentially and equally. In this paper, we propose an uncertainty-aware image captioning framework, which parallelly and iteratively operates insertion of discontinuous candidate words between existing words from easy to difficult until converged. We hypothesize that high-uncertainty words in a sentence need more prior information to make a correct decision and should be produced at a later stage. The resulting non-autoregressive hierarchy makes the caption generation explainable and intuitive. Specifically, we utilize an image-conditioned bag-of-word model to measure the word uncertainty and apply a dynamic programming algorithm to construct the training pairs. During inference, we devise an uncertainty-adaptive parallel beam search technique that yields an empirically logarithmic time complexity. Extensive experiments on the MS COCO benchmark reveal that our approach outperforms the strong baseline and related methods on both captioning quality as well as decoding speed.

Masked Auto-Encoder (MAE) pretraining methods randomly mask image patches and then train a vision Transformer to reconstruct the original pixels based on the unmasked patches. While they demonstrates impressive performance for downstream vision tasks, it generally requires a large amount of training resource. In this paper, we introduce a novel Generative Adversarial Networks alike framework, referred to as GAN-MAE, where a generator is used to generate the masked patches according to the remaining visible patches, and a discriminator is employed to predict whether the patch is synthesized by the generator. We believe this capacity of distinguishing whether the image patch is predicted or original is benefit to representation learning. Another key point lies in that the parameters of the vision Transformer backbone in the generator and discriminator are shared. Extensive experiments demonstrate that adversarial training of GAN-MAE framework is more efficient and accordingly outperforms the standard MAE given the same model size, training data, and computation resource. The gains are substantially robust for different model sizes and datasets, in particular, a ViT-B model trained with GAN-MAE for 200 epochs outperforms the MAE with 1600 epochs on fine-tuning top-1 accuracy of ImageNet-1k with much less FLOPs. Besides, our approach also works well at transferring downstream tasks.

RGB-T tracking aims to leverage the mutual enhancement and complement ability of RGB and TIR modalities for improving the tracking process in various scenarios, where cross-modal interaction is the key component. Some previous methods concatenate the RGB and TIR search region features directly to perform a coarse interaction process with redundant background noises introduced. Many other methods sample candidate boxes from search frames and conduct various fusion approaches on isolated pairs of RGB and TIR boxes, which limits the cross-modal interaction within local regions and brings about inadequate context modeling. To alleviate these limitations, we propose a novel Template-Bridged Search region Interaction (TBSI) module which exploits templates as the medium to bridge the cross-modal interaction between RGB and TIR search regions by gathering and distributing target-relevant object and environment contexts. Original templates are also updated with enriched multimodal contexts from the template medium. Our TBSI module is inserted into a ViT backbone for joint feature extraction, search-template matching, and cross-modal interaction. Extensive experiments on three popular RGB-T tracking benchmarks demonstrate our method achieves new state-of-the-art performances. Code is available at https://github.com/RyanHTR/TBSI.

Panoptic narrative grounding (PNG) aims to segment things and stuff objects in an image described by noun phrases of a narrative caption. As a multimodal task, an essential aspect of PNG is the visual-linguistic interaction between image and caption. The previous two-stage method aggregates visual contexts from offline-generated mask proposals to phrase features, which tend to be noisy and fragmentary. The recent one-stage method aggregates only pixel contexts from image features to phrase features, which may incur semantic misalignment due to lacking object priors. To realize more comprehensive visual-linguistic interaction, we propose to enrich phrases with coupled pixel and object contexts by designing a Phrase-Pixel-Object Transformer Decoder (PPO-TD), where both fine-grained part details and coarse-grained entity clues are aggregated to phrase features. In addition, we also propose a Phrase-Object Contrastive Loss (POCL) to pull closer the matched phrase-object pairs and push away unmatched ones for aggregating more precise object contexts from more phrase-relevant object tokens. Extensive experiments on the PNG benchmark show our method achieves new state-of-the-art performance with large margins.

Zero-shot video recognition (ZSVR) is a task that aims to recognize video categories that have not been seen during the model training process. Recently, vision-language models (VLMs) pre-trained on large-scale image-text pairs have demonstrated impressive transferability for ZSVR. To make VLMs applicable to the video domain, existing methods often use an additional temporal learning module after the image-level encoder to learn the temporal relationships among video frames. Unfortunately, for video from unseen categories, we observe an abnormal phenomenon where the model that uses spatial-temporal feature performs much worse than the model that removes temporal learning module and uses only spatial feature. We conjecture that improper temporal modeling on video disrupts the spatial feature of the video. To verify our hypothesis, we propose Feature Factorization to retain the orthogonal temporal feature of the video and use interpolation to construct refined spatial-temporal feature. The model using appropriately refined spatial-temporal feature performs better than the one using only spatial feature, which verifies the effectiveness of the orthogonal temporal feature for the ZSVR task. Therefore, an Orthogonal Temporal Interpolation module is designed to learn a better refined spatial-temporal video feature during training. Additionally, a Matching Loss is introduced to improve the quality of the orthogonal temporal feature. We propose a model called OTI for ZSVR by employing orthogonal temporal interpolation and the matching loss based on VLMs. The ZSVR accuracies on popular video datasets (i.e., Kinetics-600, UCF101 and HMDB51) show that OTI outperforms the previous state-of-the-art method by a clear margin.Our codes are publicly available at https://github.com/yanzhu/mm2023_oti.

Food recognition plays an important role in food choice and intake, which is essential to the health and well-being of humans. It is thus of importance to the computer vision community, and can further support many food-oriented vision and multimodal tasks, e.g., food detection and segmentation, cross-modal recipe retrieval and generation. Unfortunately, we have witnessed remarkable advancements in generic visual recognition for released large-scale datasets, yet largely lags in the food domain. In this paper, we introduce Food2K, which is the largest food recognition dataset with 2,000 categories and over 1 million images. Compared with existing food recognition datasets, Food2K bypasses them in both categories and images by one order of magnitude, and thus establishes a new challenging benchmark to develop advanced models for food visual representation learning. Furthermore, we propose a deep progressive region enhancement network for food recognition, which mainly consists of two components, namely progressive local feature learning and region feature enhancement. The former adopts improved progressive training to learn diverse and complementary local features, while the latter utilizes self-attention to incorporate richer context with multiple scales into local features for further local feature enhancement. Extensive experiments on Food2K demonstrate the effectiveness of our proposed method. More importantly, we have verified better generalization ability of Food2K in various tasks, including food image recognition, food image retrieval, cross-modal recipe retrieval, food detection and segmentation. Food2K can be further explored to benefit more food-relevant tasks including emerging and more complex ones (e.g., nutritional understanding of food), and the trained models on Food2K can be expected as backbones to improve the performance of more food-relevant tasks. We also hope Food2K can serve as a large scale fine-grained visual recognition benchmark, and contributes to the development of large scale fine-grained visual analysis.

Most of the existing semantic segmentation approaches with image-level class labels as supervision, highly rely on the initial class activation map (CAM) generated from the standard classification network. In this paper, a novel “Progressive Patch Learning” approach is proposed to improve the local details extraction of the classification, producing the CAM better covering the whole object rather than only the most discriminative regions as in CAMs obtained in conventional classification models. “Patch Learning” destructs the feature maps into patches and independently processes each local patch in parallel before the final aggregation. Such a mechanism enforces the network to find weak information from the scattered discriminative local parts, achieving enhanced local details sensitivity. “Progressive Patch Learning” further extends the feature destruction and patch learning to multi-level granularities in a progressive manner. Cooperating with a multi-stage optimization strategy, such a “Progressive Patch Learning” mechanism implicitly provides the model with the feature extraction ability across different locality-granularities. As an alternative to the implicit multi-granularity progressive fusion approach, we additionally propose an explicit method to simultaneously fuse features from different granularities in a single model, further enhancing the CAM quality on the full object coverage. Our proposed method achieves outstanding performance on the PASCAL VOC 2012 dataset (e.g., with 69.6$\%$mIoU on thetestset), which surpasses most existing weakly supervised semantic segmentation methods.

Optimising the approximation of Average Precision (AP) has been widely studied for image retrieval. Limited by the definition of AP, such methods consider both negative and positive instances ranking before each positive instance. However, we claim that only penalizing negative instances before positive ones is enough, because the loss only comes from these negative instances. To this end, we propose a novel loss, namely Penalizing Negative instances before Positive ones (PNP), which can directly minimize the number of negative instances before each positive one. In addition, AP-based methods adopt a fixed and sub-optimal gradient assignment strategy. Therefore, we systematically investigate different gradient assignment solutions via constructing derivative functions of the loss, resulting in PNP-I with increasing derivative functions and PNP-D with decreasing ones. PNP-I focuses more on the hard positive instances by assigning larger gradients to them and tries to make all relevant instances closer. In contrast, PNP-D pays less attention to such instances and slowly corrects them. For most real-world data, one class usually contains several local clusters. PNP-I blindly gathers these clusters while PNP-D keeps them as they were. Therefore, PNP-D is more superior. Experiments on three standard retrieval datasets show consistent results with the above analysis. Extensive evaluations demonstrate that PNP-D achieves the state-of-the-art performance. Code is available at https://github.com/interestingzhuo/PNPloss

Panoptic Narrative Grounding (PNG) is an emerging task whose goal is to segment visual objects of things and stuff categories described by dense narrative captions of a still image. The previous two-stage approach first extracts segmentation region proposals by an off-the-shelf panoptic segmentation model, then conducts coarse region-phrase matching to ground the candidate regions for each noun phrase. However, the two-stage pipeline usually suffers from the performance limitation of low-quality proposals in the first stage and the loss of spatial details caused by region feature pooling, as well as complicated strategies designed for things and stuff categories separately. To alleviate these drawbacks, we propose a one-stage end-to-end Pixel-Phrase Matching Network (PPMN), which directly matches each phrase to its corresponding pixels instead of region proposals and outputs panoptic segmentation by simple combination. Thus, our model can exploit sufficient and finer cross-modal semantic correspondence from the supervision of densely annotated pixel-phrase pairs rather than sparse region-phrase pairs. In addition, we also propose a Language-Compatible Pixel Aggregation (LCPA) module to further enhance the discriminative ability of phrase features through multi-round refinement, which selects the most compatible pixels for each phrase to adaptively aggregate the corresponding visual context. Extensive experiments show that our method achieves new state-of-the-art performance on the PNG benchmark with 4.0 absolute Average Recall gains.

Vision transformer emerges as a potential architecture for vision tasks. However, the intense computation and non-negligible delay hinder its application in the real world. As a widespread model compression technique, existing post-training quantization methods still cause severe performance drops. We find the main reasons lie in (1) the existing calibration metric is inaccurate in measuring the quantization influence for extremely low-bit representation, and (2) the existing quantization paradigm is unfriendly to the power-law distribution of Softmax. Based on these observations, we propose a novel Accurate Post-training Quantization framework for Vision Transformer, namely APQ-ViT. We first present a unified Bottom-elimination Blockwise Calibration scheme to optimize the calibration metric to perceive the overall quantization disturbance in a blockwise manner and prioritize the crucial quantization errors that influence more on the final output. Then, we design a Matthew-effect Preserving Quantization for Softmax to maintain the power-law character and keep the function of the attention mechanism. Comprehensive experiments on large-scale classification and detection datasets demonstrate that our APQ-ViT surpasses the existing post-training quantization methods by convincing margins, especially in lower bit-width settings (e.g., averagely up to 5.17% improvement for classification and 24.43% for detection on W4A4). We also highlight that APQ-ViT enjoys versatility and works well on diverse transformer variants.

Due to the rapid growth of online video data, video-text retrieval techniques are in urgent need, which aim to search for the most relevant video given a natural language caption and vice versa. The major challenge of this task is how to identify the true fine-grained semantic correspondence between videos and texts, using only the document-level correspondence. To deal with this issue, we propose a simple yet effective two-stream framework which takes the concept information into account and introduces a new branch of semantic-level matching. We further propose a concept propagation mechanism for mining the latent semantics in videos and achieving enriched representations. The concept propagation is achieved by building a commonsense graph distilled from ConceptNet with concepts extracted from videos and captions. The original concepts of videos are detected by pretrained detectors as the initial concept representations. By conducting attentional graph reasoning on the commonsense graph with the guidance of external knowledge, we can extend some new concepts in a detector-free manner for further enriching the video representations. In addition, a propagated BCE loss is designed for supervising the concept propagation procedure. Common space learning is then constructed for cross-modal matching. We conduct extensive experiments on various baseline models and several benchmark datasets. Promising experimental results demonstrate the effectiveness and generalization ability of our method.

Image-text matching is a fundamental research topic bridging vision and language. Recent works use hard negative mining to capture the multiple correspondences between visual and textual domains. Unfortunately, the truly informative negative samples are quite sparse in the training data, which are hard to obtain only in a randomly sampled mini-batch. Motivated by causal inference, we aim to overcome this shortcoming by carefully analyzing the analogy between hard negative mining and causal effects optimizing. Further, we propose Counterfactual Matching (CFM) framework for more effective image-text correspondence mining. CFM contains three major components, \ie, Gradient-Guided Feature Selection for automatic casual factor identification, Self-Exploration for causal factor completeness, and Self-Adjustment for counterfactual sample synthesis. Compared with traditional hard negative mining, our method largely alleviates the over-fitting phenomenon and effectively captures the fine-grained correlations between image and text modality. We evaluate our CFM in combination with three state-of-the-art image-text matching architectures. Quantitative and qualitative experiments conducted on two publicly available datasets demonstrate its strong generality and effectiveness. Code is available at: https://github.com/weihao20/cfm.

Zero-shot video classification (ZSVC) that aims to recognize video classes that have never been seen during model training, has become a thriving research direction. ZSVC is achieved by building mappings between visual and semantic embeddings. Recently, ZSVC has been achieved by automatically mining the underlying objects in videos as attributes and incorporating external commonsense knowledge. However, the object mined from seen categories can not generalized to unseen ones. Besides, the category-object relationships are usually extracted from commonsense knowledge or word embedding, which is not consistent with video modality. To tackle these issues, we propose to mine associated objects and category-object relationships for each category from retrieved web images. The associated objects of all categories are employed as generic attributes and the mined category-object relationships could narrow the modality inconsistency for better knowledge transfer. Another issue of existing ZSVC methods is that the model sufficiently trained with labeled seen categories may not generalize well to distinct unseen categories. To encourage a more reliable transfer, we propose Task Similarity aware Representation Learning (TSRL). In TSRL, the similarity between seen categories and the unseen ones is estimated and used to regularize the model in an appropriate way. We construct a model for ZSVC based on the constructed attributes, the mined category-object relationships and the proposed TSRL. Experimental results on four public datasets, i.e., FCVID, UCF101, HMDB51 and Olympic Sports, show that our model performs favorably against state-of-the-art methods. Our codes are publicly available at https://github.com/junbaoZHUO/TSRL.

We explore the capability of plain Vision Transformers (ViTs) for semantic segmentation and propose the SegViT. Previous ViT-based segmentation networks usually learn a pixel-level representation from the output of the ViT. Differently, we make use of the fundamental component—attention mechanism, to generate masks for semantic segmentation. Specifically, we propose the Attention-to-Mask (ATM) module, in which the similarity maps between a set of learnable class tokens and the spatial feature maps are transferred to the segmentation masks. Experiments show that our proposed SegViT using the ATM module outperforms its counterparts using the plain ViT backbone on the ADE20K dataset and achieves new state-of-the-art performance on COCO-Stuff-10K and PASCAL-Context datasets. Furthermore, to reduce the computational cost of the ViT backbone, we propose query-based down-sampling (QD) and query-based up-sampling (QU) to build a Shrunk structure. With our Shrunk structure, the model can save up to 40% computations while maintaining competitive performance.

Generating precise class-aware pseudo ground-truths, a.k.a, class activation maps (CAMs), is essential for Weakly-Supervised Semantic Segmentation. The original CAM method usually produces incomplete and inaccurate localization maps. To tackle with this issue, this paper proposes an Expansion and Shrinkage scheme based on the offset learning in the deformable convolution, to sequentially improve the recall and precision of the located object in the two respective stages. In the Expansion stage, an offset learning branch in a deformable convolution layer, referred to as expansion sampler'', seeks to sample increasingly less discriminative object regions, driven by an inverse supervision signal that maximizes image-level classification loss. The located more complete object region in the Expansion stage is then gradually narrowed down to the final object region during the Shrinkage stage. In the Shrinkage stage, the offset learning branch of another deformable convolution layer referred to as theshrinkage sampler'', is introduced to exclude the false positive background regions attended in the Expansion stage to improve the precision of the localization maps. We conduct various experiments on PASCAL VOC 2012 and MS COCO 2014 to well demonstrate the superiority of our method over other state-of-the-art methods for Weakly-Supervised Semantic Segmentation. The code is available at https://github.com/TyroneLi/ESOL_WSSS.

We present a simple yet effective fully convolutional one-stage 3D object detector for LiDAR point clouds of autonomous driving scenes, termed FCOS-LiDAR. Unlike the dominant methods that use the bird-eye view (BEV), our proposed detector detects objects from the range view (RV, a.k.a. range image) of the LiDAR points. Due to the range view's compactness and compatibility with the LiDAR sensors' sampling process on self-driving cars, the range view-based object detector can be realized by solely exploiting the vanilla 2D convolutions, departing from the BEV-based methods which often involve complicated voxelization operations and sparse convolutions. For the first time, we show that an RV-based 3D detector with standard 2D convolutions alone can achieve comparable performance to state-of-the-art BEV-based detectors while being significantly faster and simpler. More importantly, almost all previous range view-based detectors only focus on single-frame point clouds since it is challenging to fuse multi-frame point clouds into a single range view. In this work, we tackle this challenging issue with a novel range view projection mechanism, and for the first time demonstrate the benefits of fusing multi-frame point clouds for a range-view based detector. Extensive experiments on nuScenes show the superiority of our proposed method and we believe that our work can be strong evidence that an RV-based 3D detector can compare favourably with the current mainstream BEV-based detectors. Code will be made publicly available.

Recognizing the category and its ingredient composition from food images facilitates automatic nutrition estimation, which is crucial to various health relevant applications, such as nutrition intake management and healthy diet recommendation. Since food is composed of ingredients, discovering ingredient-relevant visual regions can help identify its corresponding category and ingredients. Furthermore, various ingredient relationships like co-occurrence and exclusion are also critical for this task. For that, we propose an ingredient-oriented multi-task food category-ingredient joint learning framework for simultaneous food recognition and ingredient prediction. This framework mainly involves learning an ingredient dictionary for ingredient-relevant visual region discovery and building an ingredient-based semantic-visual graph for ingredient relationship modeling. To obtain ingredient-relevant visual regions, we build an ingredient dictionary to capture multiple ingredient regions and obtain the corresponding assignment map, and then pool the region features belonging to the same ingredient to identify the ingredients more accurately and meanwhile improve the classification performance. For ingredient-relationship modeling, we utilize the visual ingredient representations as nodes and the semantic similarity between ingredient embeddings as edges to construct an ingredient graph, and then learn their relationships via the graph convolutional network to make label embeddings and visual features interact with each other to improve the performance. Finally, fused features from both ingredient-oriented region features and ingredient-relationship features are used in the following multi-task category-ingredient joint learning. Extensive evaluation on three popular benchmark datasets (ETH Food-101, Vireo Food-172 and ISIA Food-200) demonstrates the effectiveness of our method. Further visualization of ingredient assignment maps and attention maps also shows the superiority of our method.

BiSeNet [28], [27] has been proved to be a popular two-stream network for real-time segmentation. However, its principle of adding an extra path to encode spatial information is time-consuming, and the backbones borrowed from pretrained tasks, e.g., image classification, may be inefficient for image segmentation due to the deficiency of task-specific design. To handle these problems, we propose a novel and efficient structure named Short-Term Dense Concatenate network (STDC network) by removing structure redundancy. Specifically, we gradually reduce the dimension of feature maps and use the aggregation of them for image representation, which forms the basic module of STDC network. In the decoder, we propose a Detail Aggregation module by integrating the learning of spatial information into low-level layers in single-stream manner. Finally, the low-level features and deep features are fused to predict the final segmentation results. Extensive experiments on Cityscapes and CamVid dataset demonstrate the effectiveness of our method by achieving promising trade-off between segmentation accuracy and inference speed. On Cityscapes, we achieve 71.9% mIoU on the test set with a speed of 250.4 FPS on NVIDIA GTX 1080Ti, which is 45.2% faster than the latest methods, and achieve 76.8% mIoU with 97.0 FPS while inferring on higher resolution images. Code is available at https://github.com/MichaelFan01/STDC-Seg.

Weakly Supervised Semantic Segmentation (WSSS) with image-level annotation uses class activation maps from the classifier as pseudo-labels for semantic segmentation. However, such activation maps usually highlight the local discriminative regions rather than the whole object, which deviates from the requirement of semantic segmentation. To explore more comprehensive class-specific activation maps, we propose an Embedded Discriminative Attention Mechanism (EDAM) by integrating the activation map generation into the classification network directly for WSSS. Specifically, a Discriminative Activation (DA) layer is designed to explicitly produce a series of normalized class-specific masks, which are then used to generate class-specific pixel-level pseudo-labels demanded in segmentation. For learning the pseudo-labels, the masks are multiplied with the feature maps after the backbone to generate the discriminative activation maps, each of which encodes the specific information of the corresponding category in the input images. Given such class-specific activation maps, a Collaborative Multi-Attention (CMA) module is proposed to extract the collaborative information of each given category from images in a batch. In inference, we directly use the activation masks from the DA layer as pseudo-labels for segmentation. Based on the generated pseudo-labels, we achieve the mIoU of 70.60% on PASCAL VOC 2012 segmentation testset, which is the new state-of-the-art, to our best knowledge. Code and pre-trained models are available online soon.

Open-set semi-supervised learning (open-set SSL) investigates a challenging but practical scenario where out-of-distribution (OOD) samples are contained in the unlabeled data. While the mainstream technique seeks to completely filter out the OOD samples for semi-supervised learning (SSL), we propose a novel training mechanism that could effectively exploit the presence of OOD data for enhanced feature learning while avoiding its adverse impact on the SSL. We achieve this goal by first introducing a warm-up training that leverages all the unlabeled data, including both the in-distribution (ID) and OOD samples. Specifically, we perform a pretext task that enforces our feature extractor to obtain a high-level semantic understanding of the training images, leading to more discriminative features that can benefit the downstream tasks. Since the OOD samples are inevitably detrimental to SSL, we propose a novel cross-modal matching strategy to detect OOD samples. Instead of directly applying binary classification [39], we train the network to predict whether the data sample is matched to an assigned one-hot class label. The appeal of the proposed cross-modal matching over binary classification is the ability to generate a compatible feature space that aligns with the core classification task. Extensive experiments show that our approach substantially lifts the performance on open-set SSL and outperforms the state-of-the-art by a large margin.

Face clustering plays an essential role in exploiting massive unlabeled face data. Recently, graph-based face clustering methods are getting popular for their satisfying performances. However, they usually suffer from excessive memory consumption especially on large-scale graphs, and rely on empirical thresholds to determine the connectivities between samples in inference, which restricts their applications in various real-world scenes. To address such problems, in this paper, we explore face clustering from the pairwise angle. Specifically, we formulate the face clustering task as a pairwise relationship classification task, avoiding the memory-consuming learning on large-scale graphs. The classifier can directly determine the relationship between samples and is enhanced by taking advantage of the contextual information. Moreover, to further facilitate the efficiency of our method, we propose a rank-weighted density to guide the selection of pairs sent to the classifier. Experimental results demonstrate that our method achieves state-of-the-art performances on several public clustering benchmarks at the fastest speed and shows a great advantage in comparison with graph-based clustering methods on memory consumption.

Recently, lane detection has made great progress with the rapid development of deep neural networks and autonomous driving. However, there exist three mainly problems including characterizing lanes, modeling the structural relationship between scenes and lanes, and supporting more attributes (e.g., instance and type) of lanes. In this paper, we propose a novel structure guided framework to solve these problems simultaneously. In the framework, we first introduce a new lane representation to characterize each instance. Then a top-down vanishing point guided anchoring mechanism is proposed to produce intensive anchors, which efficiently capture various lanes. Next, multi-level structural constraints are used to improve the perception of lanes. In the process, pixel-level perception with binary segmentation is introduced to promote features around anchors and restore lane details from bottom up, a lane-level relation is put forward to model structures (i.e., parallel) around lanes, and an image-level attention is used to adaptively attend different regions of the image from the perspective of scenes. With the help of structural guidance, anchors are effectively classified and regressed to obtain precise locations and shapes. Extensive experiments on public benchmark datasets show that the proposed approach outperforms state-of-the-art methods with 117 FPS on a single GPU.

Referring Image Segmentation (RIS) aims at segmenting the target object from an image referred by one given natural language expression. The diverse and flexible expressions and complex visual contents in the images raise the RIS model with higher demands for investigating fine-grained matching behaviors between words in expressions and objects presented in images. However, such matching behaviors are hard to be learned and captured when the visual cues of referents (i.e. referred objects) are insufficient, as the referents of weak visual cues tend to be easily confused by cluttered background at boundary or even overwhelmed by salient objects in the image. And the insufficient visual cues issue can not be handled by the cross-modal fusion mechanisms as done in previous work.In this paper, we tackle this problem from a novel perspective of enhancing the visual information for the referents by devising a Two-stage Visual cues enhancement Network (TV-Net), where a novel Retrieval and Enrichment Scheme (RES) and an Adaptive Multi-resolution feature Fusion (AMF) module are proposed. Specifically, RES retrieves the most relevant image from an external data pool with regard to both the visual and textual similarities, and then enriches the visual information of the referent with the retrieved image for better multimodal feature learning. AMF further enhances the visual detailed information by incorporating the high-resolution feature maps from lower convolution layers of the image. Through the two-stage enhancement, our proposed TV-Net enjoys better performances in learning fine-grained matching behaviors between the natural language expression and image, especially when the visual information of the referent is inadequate, thus produces better segmentation results. Extensive experiments are conducted to validate the effectiveness of the proposed method on the RIS task, with our proposed TV-Net surpassing the state-of-the-art approaches on four benchmark datasets.

Very recently, a variety of vision transformer architectures for dense prediction tasks have been proposed and they show that the design of spatial attention is critical to their success in these tasks. In this work, we revisit the design of the spatial attention and demonstrate that a carefully devised yet simple spatial attention mechanism performs favorably against the state-of-the-art schemes. As a result, we propose two vision transformer architectures, namely, Twins- PCPVT and Twins-SVT. Our proposed architectures are highly efficient and easy to implement, only involving matrix multiplications that are highly optimized in modern deep learning frameworks. More importantly, the proposed architectures achieve excellent performance on a wide range of visual tasks including image-level classification as well as dense detection and segmentation. The simplicity and strong performance suggest that our proposed architectures may serve as stronger backbones for many vision tasks.

Most deep learning based image inpainting approaches adopt autoencoder or its variants to fill missing regions in images. Encoders are usually utilized to learn powerful representational spaces, which are important for dealing with sophisticated learning tasks. Specifically, in image inpainting tasks, masks with any shapes can appear anywhere in images (i.e., free-form masks) which form complex patterns. It is difficult for encoders to capture such powerful representations under this complex situation. To tackle this problem, we propose a self-supervised Siamese inference network to improve the robustness and generalization. It can encode contextual semantics from full resolution images and obtain more discriminative representations. we further propose a multi-scale decoder with a novel dual attention fusion module (DAF), which can combine both the restored and known regions in a smooth way. This multi-scale architecture is benefit for decoding discriminative representations learned by encoders into images layer by layer. In this way, unknown regions will be filled naturally from outside to inside. Qualitative and quantitative experiments on multiple datasets, including facial and natural datasets (i.e., Celeb-HQ, Pairs Street View, Places2 and ImageNet), demonstrate that our proposed method outperforms state-of-the-art methods in generating high-quality inpainting results.

In recent years, scene text recognition is always regarded as a sequence-to-sequence problem. Connectionist Temporal Classification (CTC) and Attentional sequence recognition (Attn) are two very prevailing approaches to tackle this problem while they may fail in some scenarios respectively. CTC concentrates more on every individual character but is weak in text semantic dependency modeling. Attn based methods have better context semantic modeling ability while tends to overfit on limited training data. In this paper, we elaborately design a Rectified Attentional Double Supervised Network (ReADS) for general scene text recognition. To overcome the weakness of CTC and Attn, both of them are applied in our method but with different modules in two supervised branches which can make a complementary to each other. Moreover, effective spatial and channel attention mechanisms are introduced to eliminate background noise and extract valid foreground information. Finally, a simple rectified network is implemented to rectify irregular text. The ReADS can be trained end-to-end and only word-level annotations are required. Extensive experiments on various benchmarks verify the effectiveness of ReADS which achieves state-of-the-art performance.

Benefiting from the success of deep learning, Optical Character Recognition (OCR) is booming in recent years. As we all know, the text recognition results are vulnerable to slight perturbation in input images, thus a method for measuring how reliable the results are is crucial. In this paper, we present a novel method for confidence measurement given a text recognition result, which can be embedded in any text recognizer with little overheads. Our method consists of two stages with a coarse-to-fine style. The first stage generates multiple candidates for voting coarse scores by a Single-Input Multi-Output network (SIMO). The second stage calculates a refined confidence score referred by the voting result and the conditional probabilities of the Top-1 probable recognition sequence. Highly competitive performance is achieved on several standard benchmarks which validate the efficiency and effectiveness of the proposed method. Moreover, it can be adopted in both Latin and non-Latin languages.

Food recognition has received more and more attention in the multimedia community for its various real-world applications, such as diet management and self-service restaurants. A large-scale ontology of food images is urgently needed for developing advanced large-scale food recognition algorithms, as well as for providing the benchmark dataset for such algorithms. To encourage further progress in food recognition, we introduce the dataset ISIA Food-500 with 500 categories from the list in the Wikipedia and 399,726 images, a more comprehensive food dataset that surpasses existing popular benchmark datasets by category coverage and data volume. Furthermore, we propose a stacked global-local attention network, which consists of two sub-networks for food recognition. One sub-network first utilizes hybrid spatial-channel attention to extract more discriminative features, and then aggregates these multi-scale discriminative features from multiple layers into global-level representation (e.g., texture and shape information about food). The other one generates attentional regions (e.g., ingredient relevant regions) from different regions via cascaded spatial transformers, and further aggregates these multi-scale regional features from different layers into local-level representation. These two types of features are finally fused as comprehensive representation for food recognition. Extensive experiments on ISIA Food-500 and other two popular benchmark datasets demonstrate the effectiveness of our proposed method, and thus can be considered as one strong baseline. The dataset, code and models can be found at http://123.57.42.89/FoodComputing-Dataset/ISIA-Food500.html.

Video summarization aims to select representative frames to retain high-level information, which is usually solved by predicting the segment-wise importance score via a softmax function. However, softmax function suffers in retaining high-rank representations for complex visual or sequential information, which is known as the Softmax Bottleneck problem. In this paper, we propose a novel framework named Dual Mixture Attention (DMASum) model with Meta Learning for video summarization that tackles the softmax bottleneck problem, where the Mixture of Attention layer (MoA) effectively increases the model capacity by employing twice self-query attention that can capture the second-order changes in addition to the initial query-key attention, and a novel Single Frame Meta Learning rule is then introduced to achieve more generalization to small datasets with limited training sources. Furthermore, the DMASum significantly exploits both visual and sequential attention that connects local key-frame and global attention in an accumulative way. We adopt the new evaluation protocol on two public datasets, SumMe, and TVSum. Both qualitative and quantitative experiments manifest significant improvements over the state-of-the-art methods.