Cooperative Learning of Audio and Video Models from Self-Supervised Synchronization

Open paper

Benchmark	Model	Rank	Results
audio-classification-on-esc-50	AVTS	–	Top-1 Accuracy: 82.3
self-supervised-action-recognition-on-hmdb51-1	AVTS	–	Top-1 Accuracy: 61.6
self-supervised-action-recognition-on-ucf101-1	AVTS	–	3-fold Accuracy: 89.0