1 paper · 1 filter
Geuntaek Lim, Sungjune Park, Jaeyun Lee +5
The inherent complexity of video understanding makes it difficult to determine whether Video-LLM benchmark performance stems from visual perception, linguistic reasoning, or knowle…