Spatio-temporal Action Localization on UCF101-24 (split 1)
84.3mAPACAR
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| ACARinputs=V, Backbone=R-50, Sampling=8 x 42020.06 | 84.3 | — | — | |
| SlowFast R-50inputs=V, Sampling=8 x 42020.06 | 82.4 | — | — | |
| S3D-Ginputs=V+F2020.06 | 78.8 | — | — | |
| AIAinputs=V, Backbone=R-502020.06 | 78.8 | — | — | |
| Zhang et al.inputs=V, Backbone=I3D2020.06 | 77.9 | — | — | |
| I3Dinputs=V+F2020.06 | 76.3 | — | — | |
| STEPinputs=V+F, Backbone=I3D2020.06 | 75 | — | — | |
| ACTinputs=V2020.06 | 69.5 | — | — | |
| T-CNNinputs=V2020.06 | 67.3 | — | — | |
| ACTIoU threshold=0.52017.05 | — | 69.5 | 51.4 | |
| I3D + Faster R-CNNIoU threshold=0.5, Type=3D two-stream model2017.05 | — | 76.3 | 59.9 | |
| PengIoU threshold=0.5, Multi-Region (MR)=false2017.05 | — | 64.8 | — | |
| PengIoU threshold=0.5, Multi-Region (MR)=true2017.05 | — | 65.7 | 35.9 | |
| Singh et al.IoU threshold=0.52017.05 | — | — | 46.3 |