| Model | Dataset | Frames | Input | Top-1 | Top-1* | mAP | mAP* | Link |
|---|---|---|---|---|---|---|---|---|
| resnet101 | MMit | 5 | 224*224 | - | - | 58.9 | 60.7 | model |
| resnet50 | Kinetics-600 | 8 | 224*224 | 67.5 | 70.0 | - | - | model |
| Model | Dataset | Input | Top-1 | Top-1* | mAP | mAP* | Link | |
|---|---|---|---|---|---|---|---|---|
| resnet50 | MMit | 8 | 224*224 | - | - | 62.2 | 62.8 | model |
| resnet50 | MMit | 16 | 224*224 | - | - | 62.5 | 62.9 | model |
| resnet101 | MMit | 8 | 224*224 | - | - | 62.2 | 63.0 | model |
| resnet50 | Something | 8 | 224*224 | 46.0 | 47.1 | - | - | model |
| resnet101 | Kinetics-700 | 8 | 224*224 | 61.9 | 64.3 | - | - | model |
| Model | Dataset | Frames | Input | Top-1 | Top-1* | mAP | mAP* | Link |
|---|---|---|---|---|---|---|---|---|
| SlowFast101 | Kinetics-700 | 64(8 * 8) | 112*112 | - | 65.2 | - | - | model |
| SlowFast101 | MMit | 64(8 * 8) | 112*112 | - | - | 59.9 | 61.5 | model |
| SlowFast50 | Kinetics-600 | 64(8 * 8) | 112*112 | 70.0 | 77.5 | - | - | model |
| SlowFast50 | Kinetics-600 | 64(8 * 8) | 224*224 | 72.3 | 79.8 | - | - | model |
* : Means using multi crops and multi clips (3 * 10) when testing
TSM Models can refer to Github
MMit : Multi-Moments in Time