Action Recognition on HMDB51 (fine-tuning)
73.3AccuracyVideoMAE
Evaluation Results
| Method | Links | |
|---|---|---|
| VideoMAEBackbone=ViT-B, Extra data=Kinetics-400, Frames=16, Param=87M, Modality=V, Fine-tuning=true2022.03 | 73.3 | |
| GDTBackbone=R(2+1)D, Extra data=IG65M, Frames=32, Param=15M, Modality=V+A, Fine-tuning=true2022.03 | 72.8 | |
| ρBYOL_ρ=4Backbone=SlowOnly-R50, Extra data=Kinetics-400, Frames=8, Param=32M, Modality=V, Fine-tuning=true2022.03 | 72.1 | |
| CVRLBackbone=Slow-R152 (2x), Extra data=Kinetics-600, Frames=32, Param=328M, Modality=V, Fine-tuning=true2022.03 | 70.6 | |
| MMVBackbone=S3D-G, Extra data=AS+HTM, Frames=32, Param=9M, Modality=V+A+T, Fine-tuning=true2022.03 | 69.6 | |
| CVRLBackbone=SlowOnly-R50, Extra data=Kinetics-600, Frames=32, Param=32M, Modality=V, Fine-tuning=true2022.03 | 69.4 | |
| CORP_fBackbone=SlowOnly-R50, Extra data=Kinetics-400, Frames=32, Param=32M, Modality=V, Fine-tuning=true2022.03 | 68 | |
| CVRLBackbone=SlowOnly-R50, Extra data=Kinetics-400, Frames=32, Param=32M, Modality=V, Fine-tuning=true2022.03 | 67.9 | |
| ELOBackbone=R(2+1)D, Extra data=Youtube8M-2, Frames=N/A, Param=N/A, Modality=V+A, Fine-tuning=true2022.03 | 67.4 | |
| XDCBackbone=R(2+1)D, Extra data=IG65M, Frames=32, Param=15M, Modality=V+A, Fine-tuning=true2022.03 | 67.1 | |
| RSPNetBackbone=S3D-G, Extra data=Kinetics-400, Frames=64, Param=9M, Modality=V, Fine-tuning=true2022.03 | 64.7 | |
| CPDBackbone=ResNet50, Extra data=IG300k, Frames=16, Param=N/A, Modality=V+T, Fine-tuning=true2022.03 | 63.8 | |
| VideoMAEBackbone=ViT-B, Extra data=no external data, Frames=16, Param=87M, Modality=V, Fine-tuning=true2022.03 | 62.6 | |
| MIL-NCEBackbone=S3D, Extra data=HowTo100M, Frames=32, Param=9M, Modality=V+T, Fine-tuning=true2022.03 | 61 | |
| GDTBackbone=R(2+1)D, Extra data=Kinetics-400, Frames=32, Param=15M, Modality=V+A, Fine-tuning=true2022.03 | 60 | |
| Vi²CLRBackbone=S3D, Extra data=Kinetics-400, Frames=32, Param=9M, Modality=V, Fine-tuning=true2022.03 | 55.7 | |
| CoCLRBackbone=S3D-G, Extra data=Kinetics-400, Frames=32, Param=9M, Modality=V, Fine-tuning=true2022.03 | 54.6 | |
| MemDPCBackbone=R-2D3D, Extra data=Kinetics-400, Frames=40, Param=32M, Modality=V, Fine-tuning=true2022.03 | 54.5 | |
| Vi²CLRBackbone=S3D, Extra data=UCF101, Frames=32, Param=9M, Modality=V, Fine-tuning=true2022.03 | 52.9 | |
| CoCLRBackbone=S3D-G, Extra data=UCF101, Frames=32, Param=9M, Modality=V, Fine-tuning=true2022.03 | 52.1 | |
| VTHCLBackbone=SlowOnly-R50, Extra data=Kinetics-400, Frames=8, Param=32M, Modality=V, Fine-tuning=true2022.03 | 49.2 | |
| VideoMoCoBackbone=R(2+1)D, Extra data=Kinetics-400, Frames=16, Param=15M, Modality=V, Fine-tuning=true2022.03 | 49.2 | |
| SpeedNetBackbone=S3D-G, Extra data=Kinetics-400, Frames=64, Param=9M, Modality=V, Fine-tuning=true2022.03 | 48.8 | |
| XDCBackbone=R(2+1)D, Extra data=Kinetics-400, Frames=32, Param=15M, Modality=V+A, Fine-tuning=true2022.03 | 47.1 | |
| PaceBackbone=R(2+1)D, Extra data=Kinetics-400, Frames=16, Param=15M, Modality=V, Fine-tuning=true2022.03 | 36.6 | |
| VCOPBackbone=R(2+1)D, Extra data=UCF101, Frames=N/A, Param=N/A, Modality=V, Fine-tuning=true2022.03 | 30.9 | |
| OPNBackbone=VGG, Extra data=UCF101, Frames=N/A, Param=N/A, Modality=V, Fine-tuning=true2022.03 | 23.8 |