CAST: Cross-Attention in Space and Time for Video Action Recognition

Benchmark Model Rank Results
action-recognition-on-epic-kitchens-100CAST(ViT-B/16)#8Action@1: 49.3Verb@1: 72.5Noun@1: 60.9
action-recognition-in-videos-on-something-something-v2CAST(ViT-B/16)#27Top-1 Accuracy: 71.6
action-classification-on-kinetics-400CAST(ViT-B/16)#50Acc@1: 85.3