Is this you? Claim this profile to correct it, add a bio and choose the work people see first.
Claim this profileAcademic lineage
Possible advisorsa guess from early papers, not confirmed
- Xiaogang WangSuggested from co-authorship
Is this you? Claim this profile to confirm or dismiss it.
Works68 from public data
- Spindle Net: Person Re-identification with Human Body Region Guided Feature Decomposition and Fusion878
- Balanced Meta-Softmax for Long-Tailed Visual Recognition868
Balanced Softmax is presented, an elegant unbiased extension of Softmax, to accommodate the label distribution shift between training and testing, and it is demonstrated that Balanced Meta-Softmax outperforms state-of-the-art long-tailed classification solutions on both visual recognition and instance segmentation tasks.
- Efficient Attention: Attention with Linear Complexities822
A novel efficient attention mechanism equivalent to dot-product attention but with substantially less memory and computational costs is proposed, which allows more widespread and flexible integration of attention modules into a network, which leads to better accuracies.
- Spatio-Temporal Graph Transformer Networks for Pedestrian Trajectory Prediction654
STAR is presented, a Spatio-Temporal grAph tRansformer framework, which tackles trajectory prediction by only attention mechanisms, and achieves state-of-the-art performance on 5 commonly used real-world pedestrian prediction datasets.
- HydraPlus-Net: Attentive Deep Features for Pedestrian Analysis567
A new attentionbased deep neural network, named as HydraPlus-Net (HPnet), that multi-directionally feeds the multi-level attention maps to different feature layers to enrich the final feature representations for a pedestrian image.
- 366
- Person Re-identification with Deep Similarity-Guided Graph Neural Network324
This paper proposes a novel deep learning framework, named Similarity-Guided Graph Neural Network (SGGNN), which learns the edge weights with rich labels of gallery instance pairs directly, which provides relation fusion more precise information.
- Delving into Localization Errors for Monocular 3D Object Detection309
This work quantifies the impact introduced by each sub-task and found the ‘localization error’ is the vital factor in restricting monocular 3D detection, and investigates the underlying reasons behind localization errors.
- Learning Deep Neural Networks for Vehicle Re-ID with Visual-spatio-Temporal Path Proposals303
A two-stage framework that incorporates complex spatio-temporal information for effectively regularizing the re-identification results is proposed and extensive experiments and analysis show the effectiveness of the proposed method and individual components.
- 264
- Variational Relational Point Completion Network258
A variational framework, Variational Relational point Completion network (VRC-Net), with two appealing properties: Probabilistic Modeling and Relational Enhancement, which shows great generalizability and robustness on real-world point cloud scans.
- 249
- Learning Monocular Depth by Distilling Cross-Domain Stereo Networks221
This paper proposes to use the stereo matching network as a proxy to learn depth from synthetic data and use predicted stereo disparity maps for supervising the monocular depth estimation network.
- 219
- 178
- Pedestrian Behavior Understanding and Prediction with Deep Neural Networks168
A deep neural network (Behavior-CNN) is proposed to model pedestrian behaviors in crowded scenes, which has many applications in surveillance and multiple applications demonstrate the effectiveness of Behavior-CNN on pedestrian behavior modeling.
- GroupFormer: Group Activity Recognition with Clustered Spatial-Temporal Transformer162
A novel group activity recognition network termed GroupFormer captures spatial-temporal contextual information jointly to augment the individual and group representations effectively with a clustered spatial- Temporal transformer.
- HMS-Net: Hierarchical Multi-Scale Sparsity-Invariant Network for Sparse Depth Completion160
A sparsity-invariant multi-scale encoder-decoder network (HMS-Net) for handling sparse inputs and sparse feature maps is proposed, and a model withoutRGB guidance ranks 1st among all peer-reviewed methods without using RGB information, and the model with RGB guidance ranks 2nd among all RGB-guided methods.
- GTC: Guided Training of CTC towards Efficient and Accurate Scene Text Recognition158
This work proposes the guided training of CTC (GTC), where CTC model learns a better alignment and feature representations from a more powerful attentional guidance, and achieves robust and accurate prediction for both regular and irregular scene text while maintaining a fast inference speed.
- Unsupervised 3D Shape Completion through GAN Inversion150
The proposed ShapeInversion uses a GAN pre-trained on complete shapes by searching for a latent code that gives a complete shape that best reconstructs the given partial input, and is capable of incorporating the rich prior captured in a well-trained generative model.
- Deep Group-Shuffling Random Walk for Person Re-identification150
A novel group-shuffling random walk network for fully utilizing the affinity information between gallery images in both the training and testing processes and which outperforms state-of-the-art methods on the Market-1501, CUHK03, and DukeMTMC datasets by large margins.
- End-to-End Deep Kronecker-Product Matching for Person Re-identification147
A novel Kronecker Product Matching module to match feature maps of different persons in an end-to-end trainable deep neural network that outperforms state-of-the-art methods on the Market-1501, CUHK03, and DukeMTMC datasets, which demonstrates the effectiveness and generalization ability of the proposed approach.
- EcoNAS: Finding Proxies for Economical Neural Architecture Search137
A reliable proxy is presented and a hierarchical proxy strategy is formulated that spends more computations on candidate networks that are potentially more accurate, while discards unpromising ones in early stage with a fast proxy to lead to an economical evolutionary-based NAS (EcoNAS), which achieves an impressive 400×search time reduction.
- Collaborative Unsupervised Visual Representation Learning from Decentralized Data135
This work proposes a novel federated unsupervised learning framework, FedU, which outperforms training with only one party by over 5% and other methods by over 14% in linear and semi-supervised evaluation on non-IID data.
- Encoder-decoder with Multi-level Attention for 3D Human Shape and Pose Estimation112
This work proposes Multi-level Attention Encoder-Decoder Network (MAED), including a Spatial-Temporal Encoder (STE) and a Kinematic Topology Decoder (KTD) to model multi-level attentions in a unified framework.
- Unsupervised Domain Adaptive 3D Detection with Multi-Level Consistency96
A novel and unified framework, Multi-Level Consistency Network (MLC-Net), which employs a teacher-student paradigm to generate adaptive and reliable pseudo-targets and out-performs existing state-of-the-art methods on standard benchmarks.
- 88
- Pedestrian Behavior Modeling From Stationary Crowds With Applications to Intelligent Surveillance82
A novel model is proposed to model pedestrian behaviors by incorporating stationary crowd groups as a key component and is demonstrated through multiple applications, including walking path prediction, destination prediction, personality attribute classification, and abnormal event detection.
- CSG-Stump: A Learning Friendly CSG-Like Representation for Interpretable Shape Parsing81
CSG-Stump Net is presented, an unsupervised end-to-end network for learning shapes from point clouds and discovering the underlying constituent modeling primitives and operations as well, proven to be equivalent to CSG in terms of representation.
- 80
- Better Teacher Better Student: Dynamic Prior Knowledge for Knowledge Distillation71
The dynamic prior knowledge (DPK) is proposed, which integrates part of teacher's features as the prior knowledge before the feature distillation, which provides a fast solution in teacher model selection for any given model.
- 68
- BriNet: Towards Bridging the Intra-class and Inter-class Gaps in One-Shot Segmentation56
A framework to bridge the gap between information interaction between query and support images and to precisely localize the query objects is proposed, which outperforms other competitive methods and leads to a new state-of-the-art on both PASCAL VOC and MSCOCO dataset.
- NTIRE 2025 Challenge on Cross-Domain Few-Shot Object Detection: Methods and Results44
An overview of the 1st NTIRE 2025 CD-FSOD Challenge is presented, highlighting the proposed solutions and summarizing the results submitted by the participants.
- 42
- A Closer Look at the CLS Token for Cross-Domain Few-Shot Learning41
An intriguing phenomenon neglected by previous works for the CDFSL task based on ViT is found: leaving the CLS token to random initialization, instead of loading source-domain trained parameters, could consistently improve target-domain performance.
- Pyramid Fusion Transformer for Semantic Segmentation39
This study finds that the per-mask classification decoder on top of a single-scale feature is not effective enough to extract reliable probabilities or masks, and proposes a transformer-based Pyramid Fusion Transformer (PFT) for per-mask approach semantic segmentation with multi-scale features.
- Domain consistency regularization for unsupervised multi-source domain adaptive classification27
This paper proposes an end-to-end trainable network that exploits domain Consistency Regularization for unsupervised Multi-source domain Adaptive classification (CRMA), and designs an authorization strategy that assigns different authorities to domain-specific classifiers adaptively for optimal pseudo label prediction and self-training.
- MessyTable: Instance Association in Multiple Camera Views25
An interesting and challenging dataset that features a large number of scenes with messy tables captured from multiple camera views, which challenges existing methods in mining subtle appearance differences, reasoning based on contexts, and fusing appearance with geometric cues for establishing an association.
- Variational Relational Point Completion Network for Robust 3D Classification22
A dual-path architecture is proposed to enable principled probabilistic modeling across partial and complete clouds, and extensive experiments demonstrate that VRCNet outperforms state-of-the-art methods on all standard point cloud completion benchmarks.
- Federated Unsupervised Domain Adaptation for Face Recognition21
This work proposes federated unsupervised do-main adaptation for face recognition, FedFR, which jointly optimizes clustering-based domain adaptation and federated learning to elevate performance on the target domain.
- 21
- $L_0$ Regularized Stationary-Time Estimation for Crowd Analysis21
A robust algorithm for estimating how long a foreground pixel becomes stationary is proposed, formulated as an L0 optimization problem, which distinguishes among different foreground objects, which are close or overlapped in the spatio-temporal space by using a locally shared foreground codebook.
- 20
- 20
- Cheaper Pre-training Lunch: An Efficient Paradigm for Object Detection19
The efficiency and effectiveness of Montage pre-training are validated by extensive experiments on the MS-COCO dataset, where the results indicate that the models using Montages are able to achieve on-par or even better detection performances compared with the ImageNet pre- training.
- 17
- 17
- 13
- REFINE: Prediction Fusion Network for Panoptic Segmentation13
REFINE, pREdiction FusIon NEtwork for panoptic segmentation is presented, to achieve high-quality panoptIC segmentation by improving cross- task prediction fusion, and within-task prediction fusion.
- A Holistically-Guided Decoder for Deep Representation Learning With Applications to Semantic Segmentation and Object Detection13
This paper proposes one novel holistically-guided decoder which is introduced to obtain the high-resolution semantic-rich feature maps via the multi-scale features from the encoder, and implements the EfficientFCN architecture for semantic segmentation and HGD-FPN for object detection and instance segmentation.
- Factorized Attention: Self-Attention with Linear Complexities.13
A novel factorized attention (FA) module is proposed, which achieves the same expressive power as previous approaches with substantially less memory and computational consumption and democratizes self-attention to fields where the prohibitively high costs currently prevent its application.
- Probing Visual-Audio Representation for Video Highlight Detection via Hard-Pairs Guided Contrastive Learning11
This paper enlarges the discriminative power of feature embedding with a hard-pairs guided contrastive learning (HPCL) scheme and improves feature discrimination in HPCL with the effectiveness and superiority of the proposed methods compared to other state-of-the-art methods.
- 10
- 9
- StyleFlow For Content-Fixed Image to Image Translation8
StyleFlow is proposed, a new I2I translation model that consists of normalizing flows and a novel Style-Aware Normalization (SAN) module that supports both image-guided translation and multi-modal synthesis.
- 8
- Revisiting Continuity of Image Tokens for Cross-domain Few-shot Learning7
This paper proposes a simple yet effective method for CDFSL that better disrupts the continuity of image tokens, encouraging the model to rely less on large patterns and more on smaller ones, and shows the effectiveness of the method in reducing domain gaps and outperforming state-of-the-art works.
- Random Registers for Cross-Domain Few-Shot Learning6
A simple but effective approach is proposed for CDFSL to enhance the perturbation on attention maps by adding random registers on the semantic regions of image tokens, improving the effectiveness and efficiency of random registers.
- An Empirical Study of Pseudo-Labeling for Image-based 3D Object Detection5
This work conducts extensive experiments to investigate whether the pseudo-labels can provide effective supervision for the baseline models under varying settings and explores the feasibility of a cheaper alternative, i.e. pseudo-labeling, to leverage the unlabeled data.
- Decomposed Attention: Self-Attention with Linear Complexities5
This paper proposes a novel decomposed attention (DA) module with substantially less memory and computational consumption, which democ-ratizes self-attention to Elds where the prohibitively high costs have been preventing its application.
- Addressing Exacerbated Attention Sink for Source-Free Cross-Domain Few-Shot Learning3
A novel approach to dynamically re-weight tokens according to their relevance with target-domain classes during the target-domain finetuning, which explicitly suppresses the model's reliance on simple tokens and enhances the learning of hard tokens, reducing sink tokens and enhancing discriminability.
- 3
- MagnifierNet: Towards Semantic Adversary and Fusion for Person Re-identification2
This work proposes MagnifierNet, a triple-branch network which accurately mines details from whole to parts and introduces a novel loss function "Semantic Diversity Loss" to remove redundant overlaps across learned semantic representations.
- Improving CLIP Adaptation by Breaking Tail Alignment for Source-Free Cross-Domain Few-Shot Learning1
Adaptive Tail-Head Alignment (ATHA) is proposed, a novel fine-tuning strategy for CLIP that transforms the conventional uniform alignment paradigm to an adaptive alignment paradigm, with both alignment strengthening and weakening.
- MagnifierNet: Towards Semantic Regularization and Fusion for Person Re-identification.–
A novel network which accurately mines details for each semantic region and selectively fuse all semantic feature representations and introduces a novel loss function “Semantic Diversity Loss” (SD Loss) to facilitate feature diversity and improves regularization among all semantic regions.
- –
- –
Publication data from OpenAlex; citation counts are the higher of OpenAlex and Semantic Scholar, last synced 2026-10-15. One-sentence summaries under some papers are written by Semantic Scholar’s model. Citation counts may be lower than on Google Scholar, which indexes more sources.
Report an error
Wrong papers, two people merged into one, or a profile that should not be here? Tell us and we will fix or hide it.