Region-based Non-local Operation for Video Classification

Benchmark Model Rank Results
action-classification-on-kinetics-400RNL+TSM Ensemble(ResNet50, 8 + 16 frames)#126Acc@1: 77.4
action-recognition-in-videos-on-something-something-v1RNL+TSM Ensemble(R50+R101, ImageNet pretrained)#28Top 1 Accuracy: 54.1Top 5 Accuracy: 82.2
action-recognition-in-videos-on-something-something-v1RNL+TSM Ensemble(ResNet50, ImageNet pretrained)#33Top 1 Accuracy: 52.7Top 5 Accuracy: 81.5