1 paper · 1 filter
Zhengqian Wu, Ruizhe Li, Zijun Xu +3
Video question answering (VideoQA) aims to answer natural language questions according to the given videos. Although existing models perform well in the factoid VideoQA task, they…