ECCV 2026 accepted papers on Video
479 of the 2,834 papers accepted at ECCV 2026 (16.9%) are on Video, compared with 12.0% at ECCV 2024. They fall into 4 topics.
See ECCV 2026 in the interactive map
Working on something in this area? Describe your idea in scime atlas to see the closest papers by meaning and how crowded the spot has become.
Topics and papers
Video generation · 257 papers
- Live Avatar: Streaming Real-time Audio-Driven Avatar Generation with Infinite Length (58 citations)
- End-to-End Training for Autoregressive Video Diffusion via Self-Resampling (49 citations)
- LoGeR: Long-Context Geometric Reconstruction with Hybrid Memory (45 citations)
- AnyFlow: Any-Step Video Diffusion Model with On-Policy Flow Map Distillation (33 citations)
- TinyHistory: Lightweight Video History Embeddings via Two-Stage Context Learning (33 citations)
- FlowerDance: MeanFlow for Efficient and Refined 3D Dance Generation (32 citations)
- PhyGDPO: Physics-Aware Groupwise Direct Preference Optimization for Physically Consistent Text-to-Video Generation (27 citations)
- MemRoPE: Training-Free Infinite Video Generation via Evolving Memory Tokens (26 citations)
- ShotStream: Streaming Multi-Shot Video Generation for Interactive Storytelling (25 citations)
- VGGRPO: Towards World-Consistent Video Generation with 4D Latent Reward (24 citations)
- DriveVA: Video Action Models are Zero-Shot Drivers (23 citations)
- RL from Physical Feedback: Aligning Large Motion Models with Humanoid Control (23 citations)
and 245 more in the interactive map.
Video-language understanding · 121 papers
- Thinking in Streaming Video (22 citations)
- Video Streaming Thinking: VideoLLMs Can Watch and Think Simultaneously (22 citations)
- DiaDem: Advancing Dialogue Descriptions in Audiovisual Video Captioning for Multimodal Large Language Models (15 citations)
- EventMemAgent: Hierarchical Event-Centric Memory for Online Video Understanding with Adaptive Tool Use (15 citations)
- Think While Watching: Online Streaming Segment-Level Memory for Multi-Turn Video Reasoning in Multimodal Large Language Models (14 citations)
- Em-Garde: A Propose-Match Framework for Proactive Streaming Video Understanding (13 citations)
- Spatial-TTT: Streaming Visual-based Spatial Intelligence with Test-Time Training (12 citations)
- Vinci2: Providing Proactive Assistance in Continuous Egocentric Videos (11 citations)
- Decoding Children’s Gait Behavior (10 citations)
- A Benchmark and Multi-Agent System for Instruction-driven Cinematic Video Compilation (9 citations)
- Reflect-R1: Evidence-Driven Reflection for Self-Correction in Long Video Understanding (9 citations)
- CurveStream: Boosting Streaming Video Understanding in MLLMs via Curvature-Aware Hierarchical Visual Memory Management (8 citations)
and 109 more in the interactive map.
Video segmentation and tracking · 73 papers
- FreqPhys: Repurposing Implicit Physiological Frequency Prior for Robust Remote Photoplethysmography (7 citations)
- O-VAD: Industrial Video Anomaly Detection through Object-Centric Tracking and Reasoning (5 citations)
- SparkVSR: Interactive Video Super-Resolution via Sparse Keyframe Propagation (5 citations)
- Synthetic Visual Genome 2: Extracting Large-scale Spatio-Temporal Scene Graphs from Videos (4 citations)
- A First Exploration of Neuromorphic OT-CFM for Multi-Speaker VSR (3 citations)
- Bounding-Box Trajectories Matter for Video Anomaly Detection (3 citations)
- Event Stream-based Sign Language Translation: A High-Definition Benchmark Dataset and A Novel Baseline (3 citations)
- SPLIT: Training-Free AI-Generated and Partially Edited Video Detection via Spatial Patch‑Level Incoherence and Temporal Roughness (3 citations)
- There and Back Again: A Flexible-Frame Transformer for Multi-Exposure Fusion (3 citations)
- MLVC: A Multi-platform Learned Video Codec for Real-World Deployment (2 citations)
- PA-VAD: Diffusion-Based Pseudo-Only Video Anomaly Detection via Domain-Aligned Memory Updates (2 citations)
- Selective Synergistic Learning for Video Object-Centric Learning (2 citations)
and 61 more in the interactive map.
Action recognition · 28 papers
- V-JEPA 2.1: Unlocking Dense Features in Video Self-Supervised Learning (160 citations)
- LARY: A Latent Action Representation Yielding Benchmark (17 citations)
- FEEL (Force-Enhanced Egocentric Learning): A Dataset for Physical Action Understanding (7 citations)
- ReynoldsFlow: Physics-Inspired Spatiotemporal Flow Representation for Video Understanding (2 citations)
- Learning Probabilistic Embeddings for Unsupervised Action Segmentation (1 citation)
- MoAKE: Toward Unified All-in-One Action Quality Assessment via Mixture of Action Knowledge Experts (1 citation)
- OnPoint: Offline-to-Online Multi-Level Distillation for Point-Supervised Online Temporal Action Localization (1 citation)
- Adaptive Latent Trajectory Anchoring for Action Segmentation Dataset Condensation
- Amplify, Aggregate, and Adjust: VideoMAE-based Holistic-Subtle Aggregation for Micro-Action Recognition
- Causal Yet Future-Aware: Dual-Path Temporal Modeling for Online Action Segmentation
- Context-Interactive Reasoning for Group Activity Detection
- DART: Deformable Adaptive Reasoning with Temporal Queries for Online Skeleton-Based Action Recognition
and 16 more in the interactive map.
