HARK: Hierarchical Agentic Retrieval with Keyframing for Video Understanding (Student Abstract)

Current video understanding models struggle with temporal reasoning and efficient processing while balancing detail preservation with computational efficiency. We propose a hierarchical memory system that segments videos into action and scene units, combined with question-aware agentic keyframe selection. Our method achieves 70.3% overall accuracy on VideoMME short video benchmarks.

Authors

• Jingcheng Li University of California, San Diego
• Ye Qiao University of California, Irvine
• Sitao Huang University of California, Irvine

Keywords

No keywords are indexed for this paper.