Yuwei Zhan

Modularized Dynamic-Granularity Video LLM for Multi-Event Long Video Understanding

This work proposes MoD-VLLM, a novel Modularized Dynamic-Granularity Video LLM framework for multi-event long video understanding, which unifies temporal grounding and semantic understanding iteratively and self-reflectively and proposes a dynamic-granularity reinforcement learning strategy, allowing MoD-VLLM to learn optimal grounding policies and dynamic granularity visual representation jointly.