CA2ST: Cross-Attention in Audio, Space, and Time for Holistic Video Recognition.

Jongseo Lee, Joohyun Chang, Dongho Lee, Jinwoo Choi