Action Detection on JHMDB-21
85.3video-mAP@0.5CFAD
Evaluation results
| Method | Links | ||
|---|---|---|---|
| CFADBackbone=I3D, Stream Type=Two-stream, Pre-train=K6002021.04 | 85.3 | 86.8 | |
| TubeRBackbone=CSN-152, Stream Type=RGB-stream2021.04 | 82.3 | 87.4 | |
| TubeRBackbone=I3D, Stream Type=Two-stream2021.04 | 80.7 | 81.8 | |
| Li et al. 2018input=RGB+Flow2020.08 | 80.5 | 82.3 | |
| I3DBackbone=I3D, Stream Type=Two-stream2021.04 | 78.6 | — | |
| TubeRBackbone=I3D, Stream Type=RGB-stream2021.04 | 78.3 | 79.7 | |
| MOCBackbone=DLA34, Stream Type=Two-stream2021.04 | 77.2 | 77.3 | |
| Hou, Chen, and Shah 2017input=RGB+Flow2020.08 | 76.9 | 78.4 | |
| T-CNNBackbone=C3D, Stream Type=RGB-stream2021.04 | 76.9 | 78.4 | |
| 2in1Backbone=VGG, Stream Type=Two-stream2021.04 | 74.7 | — | |
| sparse-to-dense frameworkinput=RGB2020.08 | 74.3 | 76.1 | |
| Kalogeiton et al. 2017input=RGB+Flow2020.08 | 73.7 | 74.2 | |
| ACTBackbone=VGG, Stream Type=Two-stream2021.04 | 73.7 | 74.2 | |
| Song et al. 2019input=RGB+Flow2020.08 | 73.4 | 74.1 | |
| Peng and Schmid 2016input=RGB+Flow2020.08 | 73.1 | 74.3 | |
| Singh et al. 2017input=RGB+Flow2020.08 | 72 | 73.8 | |
| Saha et al. 2016input=RGB+Flow2020.08 | 71.5 | 72.6 | |
| Li et al. 2018input=RGB2020.08 | 61.7 | — | |
| Zhao and Snoek 2019input=RGB+Flow2020.08 | 58 | — | |
| Saha, Singh, and Cuzzolin 2017input=RGB2020.08 | 55.3 | 57.8 | |
| Saha et al. 2016input=RGB2020.08 | 51.3 | 52.9 |