Iterative Zoom-In: Temporal Interval Exploration for Long Video Understanding

Chenglin Li, Qianglong Chen, fengtao +1

Multimodal Large Language Models (MLLMs) have shown strong performance in video understanding tasks. However, they continue to struggle with long-form videos because of an ineffici…