Picture for Fanghong Dong

Fanghong Dong

CAST: Modeling Visual State Transitions for Consistent Video Retrieval

Add code
Mar 09, 2026
Viaarxiv icon