Action Classification on Gym99
90.6Per-Class AccuracyTQN
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| TQNPretrained dataset=K400, Modality=RGB, # frames in training=dense2021.04 | 90.6 | 74.9 | 91.6 | 93.8 | |
| LT-S3DPretrained dataset=K400, Modality=RGB, # frames in training=dense2021.04 | 88.9 | 69.1 | 90.4 | 92.5 | |
| GST-50Pretrained dataset=ImageNet, Modality=RGB, # frames in training=82021.04 | 84.6 | 53.6 | 84.9 | 89.5 | |
| TSMPretrained dataset=ImageNet, Modality=two-stream, # frames in training=32021.04 | 81.2 | 44.8 | 84.9 | 88.4 | |
| TRNmsPretrained dataset=ImageNet, Modality=two-stream, # frames in training=32021.04 | 80.2 | 47.3 | 84.9 | 87.8 | |
| TSNPretrained dataset=ImageNet, Modality=two-stream, # frames in training=32021.04 | 79.8 | 47.5 | 84.6 | 86 | |
| ST-S3DPretrained dataset=K400, Modality=RGB, # frames in training=82021.04 | 72.9 | 45.3 | 82.8 | 81.5 | |
| I3DPretrained dataset=K400, Modality=RGB, # frames in training=82021.04 | 64.4 | 47.8 | 60.2 | 75.6 |