Action Classification on UCF101 1.0 (fine-tuned)
97.3AccuracyRSPNet (Ours)
Evaluation Results
| Method | Links | |
|---|---|---|
| RSPNet (Ours)NxW/H=16x112, Backbone=R21D-18, Pre-training=30k subset of K4002023.06 | 97.3 | |
| GDTNxW/H=32x112, Backbone=R21D, Pre-training=Different pre-training data2023.06 | 95.2 | |
| CVRLNxW/H=32x224, Backbone=R21D-18, Pre-training=K4002023.06 | 92.9 | |
| VIMPACNxW/H=10x256, Backbone=ViT-L, Pre-training=Different pre-training data2023.06 | 92.7 | |
| VideoMAENxW/H=16x224, Backbone=ViT-B, Pre-training=K4002023.06 | 91.3 | |
| TCLRNxW/H=16x112, Backbone=R21D-18, Pre-training=K4002023.06 | 88.2 | |
| TransRankNxW/H=16x112, Backbone=R21D-18, Pre-training=K4002023.06 | 87.8 | |
| XDCNxW/H=32x224, Backbone=R21D, Pre-training=K4002023.06 | 84.2 | |
| AVTSNxW/H=25x224, Backbone=I3D, Pre-training=K4002023.06 | 83.7 | |
| TaCoNxW/H=16x224, Backbone=R21D-18, Pre-training=K4002023.06 | 81.8 | |
| RSPNetNxW/H=16x112, Backbone=R21D-18, Pre-training=K4002023.06 | 81.1 | |
| TempTransNxW/H=16x112, Backbone=R3D-18, Pre-training=K4002023.06 | 79.3 | |
| VideoMoCoNxW/H=16x112, Backbone=R21D-18, Pre-training=K4002023.06 | 78.7 | |
| STSNxW/H=16x112, Backbone=R21D-18, Pre-training=K4002023.06 | 77.8 | |
| PacePredNxW/H=16x112, Backbone=R21D-18, Pre-training=K4002023.06 | 77.1 | |
| VideoMAE *NxW/H=16x112, Backbone=R21D-18, Pre-training=K4002023.06 | 76.2 |