| Benchmark | Model | Rank | Results |
|---|---|---|---|
| action-recognition-in-videos-on-ucf101 | HATNet (32 frames) | #8 | 3-fold Accuracy: 97.8 |
| action-recognition-in-videos-on-hmdb-51 | HATNet (32 frames) | #23 | Average accuracy of 3 splits: 76.5 |
| action-classification-on-kinetics-400 | HATNet (32 frames) | #131 | Acc@1: 77.6 |