1 paper · 1 filter
Ziyang Wang, Honglu Zhou, Shijie Wang +6
Long video understanding (LVU) is challenging because answering real-world queries often depends on sparse, temporally dispersed cues buried in hours of mostly redundant and irrele…