ReWind: Understanding Long Videos with Instructed Learnable Memory

Anxhelo Diko, Tinghuai Wang, Wassim Swaileh +2

Vision-Language Models (VLMs) are crucial for applications requiring integrated understanding textual and visual information. However, existing VLMs struggle with long videos due t…