Computer Vision

OmniMMI: A Comprehensive Multi-modal Interaction Benchmark in Streaming Video Contexts

InteractAnything: Zero-shot Human Object Interaction Synthesis via LLM Feedback and Object Affordance Parsing

Dynamic Motion Blending for Versatile Motion Editing

MOVIS: Enhancing Multi-Object Novel View Synthesis for Indoor Scenes

Decompositional Neural Scene Reconstruction with Generative Diffusion Prior

Unveiling the Mist over 3D Vision-Language Understanding: Object-centric Evaluation with Chain-of-Analysis

GROVE: A Generalized Reward for Learning Open-Vocabulary Physical Skill

ManipTrans: Efficient Dexterous Bimanual Manipulation Transfer via Residual Learning

METASCENES: Towards Automated Replica Creation for Real-world 3D Scans

Masked Point-Entity Contrast for Open-Vocabulary 3D Scene Understanding