BEVT: BERT Pretraining of Video Transformers

Benchmark Model Rank Results
action-recognition-on-diving-48BEVT#6Accuracy: 86.7
action-recognition-in-videos-on-something-something-v2BEVT (IN-1K + Kinetics400 pretrain)#28Top-1 Accuracy: 71.4Parameters: 89GFLOPs: 321x3