Dense-Caption Matching and Frame-Selection Gating for Temporal Localization in VideoQA

37567

Knowledge Triples

Source Entity	Relationship	Target Entity
Video Question Answering Task	employs	Dual Level Attention Method
Video Question Answering Task	employs	Multi-Head Self/Cross Integration Method
Video Question Answering Task	evaluatedIn	TVQA Material
Video Question Answering Task	uses	Dense Image Captions Material