Multiview Transformers for Video Recognition

Benchmark Model Rank Results
action-classification-on-kinetics-400MTV-H (WTS 60M)#11Acc@1: 89.9Acc@5: 98.3FLOPs (G) x views: 735700x4x3
action-classification-on-kinetics-600MTV-H (WTS 60M)#8Top-1 Accuracy: 90.3Top-5 Accuracy: 98.5
action-classification-on-kinetics-700MTV-H (WTS 60M)#6Top-1 Accuracy: 83.4Top-5 Accuracy: 96.2
action-classification-on-moments-in-timeMTV-H (WTS 60M)#4Top 1 Accuracy: 47.2Top 5 Accuracy: 75.7
action-recognition-in-videos-on-somethingMTV-B#45Top-1 Accuracy: 68.5Top-5 Accuracy: 90.4
action-recognition-on-epic-kitchens-100MTV-B (WTS 60M)#6Action@1: 50.5Verb@1: 69.9Noun@1: 63.9