Is this you? Claim this profile to correct it, add a bio and choose the work people see first.
Claim this profileWorks102 from public data
- Unsupervised Person Re-identification668
A progressive unsupervised learning (PUL) method to transfer pretrained deep representations to unseen domains and demonstrates that PUL outputs discriminative features that improve the re-ID accuracy.
- 232
- PSTNet: Point Spatio-Temporal Convolution on Point Cloud Sequences168
The proposed PST convolution first disentangles space and time in point cloud sequences, then a spatial convolution is employed to capture the local structure of points in the 3D space, and a temporal Convolution is used to model the dynamics of the spatial regions along the time dimension.
- Adaptive Exploration for Unsupervised Person Re-identification154
An Adaptive Exploration (AE) method to address the domain-shift problem for re-ID in an unsupervised manner and integrates a balance strategy into the adaptive selection.
- 146
- 124
- Watching a Small Portion could be as Good as Watching All: Towards Efficient Video Classification91
An end-to-end deep reinforcement approach which enables an agent to classify videos by watching a very small portion of frames by incorporating an adaptive stop network to measure confidence score and generate timely trigger to stop the agent watching videos, which improves efficiency without loss of accuracy.
- Cubic LSTMs for Video Prediction90
Experiment shows that CubicRNN produces more accurate video predictions than prior methods on both synthetic and real-world datasets.
- 81
- 81
- Prototypical Calibrating Ambiguous Samples for Micro-Action Recognition76
A novel Prototypical Calibrating Ambiguous Network (PCAN) is proposed to unleash and mitigate the ambiguity of MAR and proposes a prototype-guided rectification to rectify prediction by incorporating the representability of prototypes.
- VideoGrain: Modulating Space-Time Attention for Multi-grained Video Editing70
VideoGrain is presented, a zero-shot approach that modulates space-time (cross- and self-) attention mechanisms to achieve fine-grained control over video content and improves feature separation by increasing intra-region awareness and reducing inter-region interference in self-attention.
- Uncovering what, why and How: A Comprehensive Benchmark for Causation Understanding of Video Anomaly70
MMEval is introduced, a novel evaluation metric designed to better align with human preferences for CUVA, facilitating the measurement of existing LLMs in comprehending the underlying cause and corresponding effect of video anoma-lies.
- Attract or Distract: Exploit the Margin of Open Set67
This paper exploits the semantic structure of open set data from two aspects: 1)Semantic Categorical Alignment, which aims to achieve good separability of target known classes by categorically aligning the centroid of target with the source, and 2) Semantic Contrastive Mapping,Which aims to push the unknown class away from the decision boundary.
- 64
- 64
- STPrivacy: Spatio-Temporal Privacy-Preserving Action Recognition53
A novel framework STPrivacy is proposed to perform video-level PPAR and introduces vision Transformers into PPAR by treating a video as a tubelet sequence, and accordingly design two complementary mechanisms, i.e., sparsification and anonymization, to remove privacy from a spatio-temporal perspective.
- Recurrent Attention Network with Reinforced Generator for Visual Dialog51
A dialog network is used to memorize the temporal context and an attention processor to parse the spatial context in Visual Dialog to reinforce G at the sentence level using the discriminative model (D), which aims to select the right answer from a few candidates to ameliorate the problem.
- 51
- Cascaded Revision Network for Novel Object Captioning42
This paper proposes a novel network structure, Cascaded Revision Network, which aims at relieving the problem of inaccurate captioning by equipping the model with out-of-domain knowledge and can accurately describe images with unseen objects.
- Motion Matters: Motion-guided Modulation Network for Skeleton-based Micro-Action Recognition37
A novel Motion-guided Modulation Network (MMN) is presented that implicitly captures and modulates subtle motion cues to enhance spatial-temporal representation learning and achieves state-of-the-art performance in skeleton-based micro-action recognition.
- 37
- EnergyMogen: Compositional Human Motion Generation with Energy-Based Diffusion Model in Latent Space36
This work proposes EnergyMoGen, which includes two spectrums of Energy-Based Models, and introduces Synergistic Energy Fusion, a semantic-aware energy model based on cross-attention, which enables semantic composition and adaptive gradient descent for text embeddings.
- Point Cloud Domain Adaptation via Masked Local 3D Structure Prediction35
This paper proposes a Masked Local Structure Prediction method to encode target data and enables models to embed source and target data in a shared feature space through estimating point cardinality, position and normal.
- DreamDPO: Aligning Text-to-3D Generation with Human Preferences via Direct Preference Optimization33
This paper proposes DreamDPO, an optimization-based framework that integrates human preferences into the 3D generation process, through direct preference optimization, and achieves competitive results, and provides higher-quality and more controllable 3D content compared to existing methods.
- Masked Spatio-Temporal Structure Prediction for Self-supervised Learning on Point Cloud Videos33
MaST-Pre is a Masked Spatio-Temporal Structure Prediction method to capture the structure of point cloud videos without human annotations and consists of two self-supervised learning tasks that are forced to model the spatial and temporal structure in point cloud videos.
- Drafting and Revision: Advancing High-Fidelity Video Inpainting31
A Drafting-and-Revision Completion Network is designed that utilizes the temporal information to complete the low-frequency semantic structure at low resolution, and a Revision Network is developed to hallucinate high-frequency details at high resolution by using the output of Drafting Network.
- TSGS: Improving Gaussian Splatting for Transparent Surface Reconstruction via Normal and De-lighting Priors28
Transparent Surface Gaussian Splatting (TSGS) is introduced, a new framework that separates geometry learning from appearance refinement and achieves accurate geometric reconstruction and realistic rendering of transparent objects simultaneously within the efficient 3DGS framework.
- 28
- Point Contrastive Prediction with Semantic Clustering for Self-Supervised Learning on Point Cloud Videos27
This paper proposes a selection strategy to retain proper negatives and make use of high-similarity samples from other instances as positive supplements for contrastive learning at the point level, and introduces a new pretext task by achieving semantic alignment of superpoints.
- Prototype Learning for Micro-gesture Classification26
The solution developed by the HFUT-VUT team utilizes two innovative modules, i.e., the cross-modal fusion module and prototypical refinement module, to improve the discriminative ability of MG features, thereby improving the classification accuracy.
- 24
- Person Tube Retrieval via Language Description24
Experimental results on person tube retrieval via language description and other two related tasks demonstrate the efficacy of Multi-Scale Structure Preservation (MSSP).
- FlowZero: Zero-Shot Text-to-Video Synthesis with LLM-Driven Dynamic Scene Syntax21
By using spatio-temporal syntaxes to guide the diffusion process, FlowZero achieves improvement in zero-shot video synthesis, generating coherent videos with vivid motion.
- 20
- 20
- TOPA: Extending Large Language Models for Video Understanding via Text-Only Pre-Alignment19
Text-Only Pre-Alignment (TOPA) is introduced, a novel approach to extend large language models (LLMs) for video understanding, without the need for pre-training on real video data, and proves competitive with recent GPT-3.5-based video agents.
- Keyword-Aware Relative Spatio-Temporal Graph Networks for Video Question Answering19
A Keyword-aware Relative Spatio-Temporal graph network for VideoQA that employs an attention mechanism to assign high weights to keywords during question encoding and integrates the relative relation modeling to better capture the spatio-temporal dynamics among object nodes.
- 19
- Informedia @ TRECVID 2016.18
This report reports on the system used in the TRECVID 2016 Multimedia Event Detection (MED) and Ad-hoc Video Search (AVS) tasks, which is a fully-automatic system with no annotation condition.
- EVA: Zero-shot Accurate Attributes and Multi-Object Video Editing17
Benefiting from the precise attention weight distribution, EVA can be easily generalized to multi-object editing scenarios and achieves accurate identity mapping.
- Variational Rectification Inference for Learning with Noisy Labels16
Variational rectification inference (VRI) is proposed to formulate the adaptive rectification for loss functions as an amortized variational inference problem and derive the evidence lower bound under the meta-learning framework, which can significantly improve the generalization performance.
- 16
- InfiniDreamer: Arbitrarily Long Human Motion Generation Via Segment Score Distillation14
Segment Score Distillation (SSD) is introduced—an optimization-based approach that leverages a pre-trained motion diffusion model trained solely on short clips to produce coherent, diverse, and context-aware long-range motions without requiring additional long-sequence training.
- Clustering for Protein Representation Learning14
A neural clustering framework that can automatically discover the critical components of a protein by considering both its primary and tertiary structure information and achieves state-of-the-art performance.
- 13
- Seeing Is Believing? A Benchmark for Multimodal Large Language Models on Visual Illusions and Anomalies12
VIA-Bench is introduced, a challenging benchmark designed to probe model performance on visual illusions and anomalies, and reveals a fundamental divergence between machine and human perception, suggesting that resolving such perceptual bottlenecks is critical for the advancement of artificial general intelligence.
- DLVINet: Advancing Dual-Lens Video Inpainting Beyond Parallax Constraints12
A sparse spatial-temporal transformer (SSTT) that effectively utilizes the information from distant frames to complete the video contents of each lens individually and a cross-lens texture transformer (CLTT) to model inter-lens consistency.
- Building Category Graphs Representation with Spatial and Temporal Attention for Visual Navigation12
A Category Relation Graph (CRG) to learn the knowledge of object category layout relations and a Temporal-Spatial-Region attention (TSR) architecture to perceive the long-term spatial-temporal dependencies of objects, aiding navigation are proposed.
- AnchorFlow: Training-Free 3D Editing via Latent Anchor-Aligned Flows11
AnchorFlow establishes a global latent anchor shared between the source and target trajectories, and enforces coherence using a relaxed anchor-alignment loss together with an anchor-aligned update rule, which ensures that transformations remain stable and semantically faithful throughout the editing process.
- Zero-1-to-A: Zero-Shot One Image to Animatable Head Avatars Using Video Diffusion11
Zero-1-to-A is a robust method that synthesizes a spatial and temporal consistency dataset for 4D avatar reconstruction using the video diffusion model and improves fidelity, animation quality, and rendering speed compared to existing diffusion-based methods, providing a solution for lifelike avatar creation.
- Can We Solve 3D Vision Tasks Starting from A 2D Vision Transformer?11
To build a 3D ViT from its 2D sibling, the patch embedding and token sequence is inflated, accompanied with new positional encoding mechanisms designed to match the 3D data geometry, and the resultant, named Simple3D-Former, performs surprisingly robustly on popular 3D tasks such as object classification, point cloud segmentation and indoor scene detection.
- 10
- Multiple kernel visual-auditory representation learning for retrieval10
This paper proposes a multiple kernel visual-auditory representation learning approach, which learns cross- media correlations from visual and auditory feature spaces with multiple kernel strategies, and gives cross-media distance measure for image-audio retrieval in the mutual subspace of co-occurrence.
- ProtChatGPT: Towards Understanding Proteins with Hybrid Representation and Large Language Models9
ProtChatGPT is introduced, which aims to learn and understand protein structures using natural language and can produce promising responses to proteins and the corresponding user questions, and could form the basis for further exploration and application in protein research.
- 9
- 9
- Progressive Point Cloud Denoising with Cross-Stage Cross-Coder Adaptive Edge Graph Convolution Network9
This paper proposes a Cross-stage Cross-coder Adaptive Edge Graph Convolution Network (C2AENet) to denoise point clouds and introduces a geometric graph and adaptively calculate edge attention based on the local and global structural information of the points.
- Prompt-Aware Adapter: Towards Learning Adaptive Visual Tokens for Multimodal Large Language Models9
This work proposes prompt-aware adapters, designed with the capability to dynamically embed visual inputs based on the specific focus of the prompt, which significantly enhances the ability of LLMs to understand and interpret visual content.
- 4DPC$^2$hat: Towards Dynamic Point Cloud Understanding with Failure-Aware Bootstrapping7
This work introduces a Mamba-enhanced temporal reasoning MLLM to capture long-range dependencies and dynamic patterns among a point cloud sequence, and proposes a failure-aware bootstrapping learning strategy that iteratively identifies model deficiencies and generates targeted QA supervision to continuously strengthen corresponding reasoning capabilities.
- SAMControl: Controlling Pose and Object for Image Editing with Soft Attention Mask7
This article proposes a method called SAMControl (Soft Attention Mask) to adaptively control the pose and object details for image editing and utilizes a direct inversion technique to ensure the fidelity of source details within SAM.
- 5
- Motion = Video - Content: Towards Unsupervised Learning of Motion Representation from Videos5
This paper designs an encoder-decoder model which consists of a content encoder, a motion encoder and a video generator and demonstrates that the proposed framework can be used for conditional video generation and fine-grained action recognition.
- ITS3D: Inference-Time Scaling for Text-Guided 3D Diffusion Models4
ITS3D, a framework that formulates the task as an optimization problem to identify the most effective Gaussian noise input, is introduced, driven by a verifier-guided search algorithm, where the search algorithm iteratively refines noise candidates based on verifier feedback.
- 4
- 4
- GraphTARIF: Linear Graph Transformer with Augmented Rank and Improved Focus3
This work proposes a novel hybrid framework that enhances linear attention by attaching a gated local graph network branch to the value matrix, thereby increasing the rank of the resulting attention map and introduces a learnable log-power function into the attention scores to reduce entropy and sharpen focus.
- Protein Captioning: Bridging the Gap between Protein Sequences and Natural Languages3
This work proposes a simple yet effective method, Protein-to-Text Generative Pre-Trained Transformer (P2T-GPT), to fuse multimodal embeddings and translate the chain of amino acid residues in a protein to a sequence of natural language words, i.e., text.
- Prior-Free Continual Learning with Unlabeled Data in the Wild3
This work proposes a Prior-Free Continual Learning (PFCL) method, which learns new tasks without knowing the task identity or any previous data, and develops a reliable sample selection strategy to obtain consistent performance improvement.
- DPMix: Mixture of Depth and Point Cloud Video Experts for 4D Action Segmentation3
The proposed method, named Mixture of Depth and Point cloud video experts (DPMix), achieved the first place in the 4D Action Segmentation Track of the HOI4D Challenge 2023 and the accuracy is significantly improved.
- TransNormal: Dense Visual Semantics for Diffusion-based Transparent Object Normal Estimation2
This work proposes TransNormal, a novel framework that adapts pre-trained diffusion priors for single-step normal regression to handle the lack of texture in transparent surfaces, and introduces TransNormal-Synthetic, a physics-based dataset with high-fidelity normal maps for transparent labware.
- 2
- Stroke3D: Lifting 2D strokes into rigged 3D model via latent diffusion models2
This work introduces Stroke3D, a novel framework that directly generates rigged meshes from user inputs: 2D drawn strokes and a descriptive text prompt, and is the first to generate rigged 3D meshes conditioned on user-drawn 2D strokes.
- 2
- Text-to-Motion Generation With Discrete Representations and Large Language Models1
This work investigates a simple yet effective conditional generative framework for text-to-motion generation and proposes T2M-GIT+, which employs a non-autoregressive method to generate discrete motion representations in parallel, and is therefore more efficient than T2M-GPT+ while achieving comparable results.
- 1
- 1
- TV-Dialogue: Crafting Theme-Aware Video Dialogues with Immersive Interaction1
This paper proposes TV-Dialogue, a novel multi-modal agent framework that ensures both theme alignment and visual consistency by enabling real-time immersive interactions among video characters, thereby accurately understanding the video content and generating new dialogue that aligns with the given themes.
- CktGen: Automated Analog Circuit Design with Generative Artificial Intelligence1
This work proposes CktGen, a simple yet effective variational autoencoder that maps discretized specifications and circuits into a joint latent space and reconstructs the circuit from that latent vector by canonicalizing the latent space with respect to specifications.
- Text-Scene: A Scene-to-Language Parsing Framework for 3D Scene Understanding1
Text-Scene is introduced, a framework that automatically parses 3D scenes into textual descriptions for scene understanding, aiming to make 3D scene content understandable through language.
- Bidirectional transition consistency between multi-domain observations for visual reinforcement learning generalization1
The Multi-Domain Bidirectional Transition model, which explicitly incorporates multi-domain observations with visual perturbations and kernel regions, enables MDBT to remove noise interference and preserve task relevance, resulting in more robust and transferable representations.
- 1
- A Study on Differentiable Logic and LLMs for EPIC-KITCHENS-100 Unsupervised Domain Adaptation Challenge for Action Recognition 20231
The innovative application of a differentiable logic loss in the training to leverage the co-occurrence relations between verb and noun, as well as the pre-trained Large Language Models to generate the logic rules for the adaptation to unseen action labels achieves the first place in terms of top-1 action recognition accuracy.
- 1
- –
- –
- –
- RegionSLM: Region-aware Question Answering on Document Screenshots–
This work presents RegionSLM, a region-aware SLM designed to explicitly connect the question to its supporting regions, and curates ReDoc, a region-supervised corpus with 105k documents and 350k question-answer pairs obtained via a question-guided two-step filtering procedure.
- –
- –
- DRFormer: A Dual-Regularized Bidirectional Transformer for Person Re-identification–
This paper analyzes the complementary roles of these two architectures and proposes a framework to synergize their strengths by a \textbf{D}ual-\textbf{R}egularized Bidirectional \textbf{Transformer} (\textbf{DRFormer}).
- –
- –
- –
- –
- –
- DeMoGen: Towards Decompositional Human Motion Generation with Energy-Based Diffusion Models–
This paper proposes DeMoGen, a compositional training paradigm for decompositional learning that employs an energy-based diffusion model that directly captures the composed distribution of multiple motion concepts, enabling the model to discover them without relying on ground-truth motions for individual concepts.
- –
- Learning Robust Representations via Bidirectional Transition for Visual Reinforcement Learning–
A Bidirectional Transition (BT) framework for representation learning is introduced that employs the bidirectional prediction of both forward and backward environmental transitions as auxiliary tasks to extract reliable representations and exhibits robust generalization performance for visual reinforcement learning tasks.
- –
- From Video Classification to Video Prediction: Deep Learning Approaches to Video Modelling–
A selective multi-instance learning method is proposed to automatically detect whether an event of interest happens in temporally untrimmed videos which usually consist of multiple video shots, and a series of point spatio-temporal correlation recurrent neural networks are proposed to jointly aggregate point coordinates and features in input, states and output.
- –
Publication data from OpenAlex, with missing venues and authors filled in from Crossref; citation counts are the higher of OpenAlex and Semantic Scholar, last synced 2026-10-12. 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.