CVPR 2025 accepted papers on Video
391 of the 2,871 papers accepted at CVPR 2025 (13.6%) are on Video, compared with 10.0% at CVPR 2024. They fall into 4 topics.
See CVPR 2025 in the interactive map
Working on something in this area? Describe your idea in scime atlas to see the closest accepted papers by meaning and how crowded the spot has become.
Topics and papers
Video generation · 157 papers
- DepthCrafter: Generating Consistent Long Depth Sequences for Open-world Videos (51 citations)
- From Slow Bidirectional to Fast Autoregressive Video Diffusion Models (37 citations)
- DriveDreamer4D: World Models Are Effective Data Machines for 4D Driving Scene Representation (35 citations)
- StreamingT2V: Consistent, Dynamic, and Extendable Long Video Generation from Text (34 citations)
- Motion Prompting: Controlling Video Generation with Motion Trajectories (32 citations)
- Hallo3: Highly Dynamic and Realistic Portrait Image Animation with Video Diffusion Transformer (29 citations)
- MMAudio: Taming Multimodal Joint Training for High-Quality Video-to-Audio Synthesis (29 citations)
- Learning Temporally Consistent Video Depth from Video Diffusion Priors (28 citations)
- AC3D: Analyzing and Improving 3D Camera Control in Video Diffusion Transformers (26 citations)
- Rethinking Diffusion for Text-Driven Human Motion Generation: Redundant Representations, Evaluation, and Masked Autoregression (24 citations)
- Go-with-the-Flow: Motion-Controllable Video Diffusion Models Using Real-Time Warped Noise (23 citations)
- Tora: Trajectory-oriented Diffusion Transformer for Video Generation (20 citations)
and 145 more in the interactive map.
Video-language understanding · 111 papers
- Video-MME: The First-Ever Comprehensive Evaluation Benchmark of Multi-modal LLMs in Video Analysis (149 citations)
- VideoTree: Adaptive Tree-based Video Representation for LLM Reasoning on Long Videos (46 citations)
- Adaptive Keyframe Sampling for Long Video Understanding (42 citations)
- Video-XL: Extra-Long Vision Language Model for Hour-Scale Video Understanding (33 citations)
- Holmes-VAU: Towards Long-term Video Anomaly Understanding at Any Granularity (31 citations)
- MLVU: Benchmarking Multi-task Long Video Understanding (30 citations)
- M-LLM Based Video Frame Selection for Efficient Video Understanding (28 citations)
- Video Summarization with Large Language Models (23 citations)
- EgoLife: Towards Egocentric Life Assistant (22 citations)
- Dispider: Enabling Video LLMs with Active Real-Time Interaction via Disentangled Perception, Decision, and Reaction (20 citations)
- The Devil is in Temporal Token: High Quality Video Reasoning Segmentation (20 citations)
- GLUS: Global-Local Reasoning Unified into A Single Large Language Model for Video Segmentation (19 citations)
and 99 more in the interactive map.
Video segmentation and tracking · 108 papers
- Towards Practical Real-Time Neural Video Compression (61 citations)
- Video Depth Anything: Consistent Depth Estimation for Super-Long Videos (61 citations)
- Change3D: Revisiting Change Detection and Captioning from A Video Modeling Perspective (31 citations)
- FineVQ: Fine-Grained User Generated Content Video Quality Assessment (29 citations)
- Generalizing Deepfake Video Detection with Plug-and-Play: Video-Level Blending and Spatiotemporal Adapter Tuning (28 citations)
- Event-based Video Super-Resolution via State Space Models (27 citations)
- Towards More General Video-based Deepfake Detection through Facial Component Guided Adaptation for Foundation Model (23 citations)
- Align3R: Aligned Monocular Depth Estimation for Dynamic Videos (22 citations)
- VideoGigaGAN: Towards Detail-rich Video Super-Resolution (22 citations)
- Discrete to Continuous: Generating Smooth Transition Poses from Sign Language Observations (18 citations)
- AG-VPReID: A Challenging Large-Scale Benchmark for Aerial-Ground Video-based Person Re-Identification (15 citations)
- Anomize: Better Open Vocabulary Video Anomaly Detection (15 citations)
and 96 more in the interactive map.
Action recognition · 15 papers
- Revealing Key Details to See Differences: A Novel Prototypical Perspective for Skeleton-based Action Recognition (36 citations)
- HD-EPIC: A Highly-Detailed Egocentric Video Dataset (19 citations)
- Weakly Supervised Temporal Action Localization via Dual-Prior Collaborative Learning Guided by Multimodal Large Language Models (16 citations)
- Context-Enhanced Memory-Refined Transformer for Online Action Detection (10 citations)
- Temporal Alignment-Free Video Matching for Few-shot Action Recognition (9 citations)
- Action Detail Matters: Refining Video Recognition with Local Action Queries (4 citations)
- GraphMimic: Graph-to-Graphs Generative Modeling from Videos for Policy Learning (4 citations)
- Understanding Multi-Task Activities from Single-Task Videos (4 citations)
- Beyond Image Classification: A Video Benchmark and Dual-Branch Hybrid Discrimination Framework for Compositional Zero-Shot Learning (3 citations)
- Boosting Point-Supervised Temporal Action Localization through Integrating Query Reformation and Optimal Transport (2 citations)
- Condensing Action Segmentation Datasets via Generative Network Inversion (2 citations)
- H-MoRe: Learning Human-centric Motion Representation for Action Analysis (2 citations)
and 3 more in the interactive map.