Institution

Central South University

A comprehensive national research university in Changsha, China, with active computer vision and machine learning groups.

Multimodal Models · Ant Group

MemDreamer: Hierarchical Graph Memory for Long Video Understanding

MemDreamer turns hours-long video QA into agentic retrieval over a 3-tier graph memory, lifting LVBench from 78.2 to 90.7 (+12.5) while the reasoning model reads ~6K tokens instead of 240K-784K.