Large Vision-Language Models (LVLMs) demonstrate remarkable performance in short-video tasks such as video question answering, but struggle in long-video understanding. The linear frame sampling ...
Some results have been hidden because they may be inaccessible to you
Show inaccessible results