Action Recognition In Videos On Ucf101 2
評価指標
3-fold Accuracy
評価結果
このベンチマークにおける各モデルのパフォーマンス結果
モデル名 | 3-fold Accuracy | Paper Title | Repository |
|---|---|---|---|
| 3D-MobileNetV2 0.2x | 55.56 | Resource Efficient 3D Convolutional Neural Networks | |
| 3D-ShuffleNetV2 0.25x | 56.52 | Resource Efficient 3D Convolutional Neural Networks | |
| Baseline UCF101 | 43.9 | UCF101: A Dataset of 101 Human Actions Classes From Videos in The Wild | |
| 3D-SqueezeNet | 74.94 | Resource Efficient 3D Convolutional Neural Networks | |
| STM (ImageNet+Kinetics pretrain) | 96.2 | STM: SpatioTemporal and Motion Encoding for Action Recognition | - |
0 of 5 row(s) selected.