MOMA: Mixture-of-Modality-Adaptations for Transferring Knowledge from Image Models Towards Efficient Audio-Visual Action Recognition.

Kai Wang, Dimitrios Hatzinakos

Browse the full ICASSP paper archive.