Table 5 Performance comparison of different methods on different datasets.
From: Basketball technique action recognition using 3D convolutional neural networks
Method | Accuracy on NTURGB + D | Accuracy on the Basketball-Action-Dataset | Accuracy on the B3D Dataset | Remarks |
|---|---|---|---|---|
Frame difference | 75.3% | 69.8% | 73.1% | Traditional method |
Optical flow | 81.4% | 76.5% | 78.9% | Traditional method |
Transformer models | 90.1% | 87.3% | 89.2% | DL method |
Reference37 | 86.2% | 82.1% | 84.3% | DL method |
Reference38 | 84.7% | 80.5% | 82.9% | DL method |
Reference39 | 88.5% | 84.9% | 86.7% | DL method |
The proposed method | 93.1% | 88.5% | 90.0% | The proposed method |
