Landmark Prediction on Human 3.6M (test)
2.16Error (Mixed Actions)Hourglass Newell et al.
Evaluation results
| Method | Links | |||||||
|---|---|---|---|---|---|---|---|---|
| Hourglass Newell et al.Supervision=Supervised2018.04 | 2.16 | 1.88 | 1.92 | 2.15 | 1.62 | 1.88 | 2.21 | |
| NewellSupervision Strategy=fully supervised baseline2021.05 | 2.16 | 1.88 | 1.92 | 2.15 | 1.62 | 1.88 | 2.21 | |
| JakabSupervision Strategy=self-supervised (unpaired labels)2021.05 | 2.73 | 2.66 | 2.27 | 2.73 | 2.35 | 2.35 | 4 | |
| LorenzSupervision Strategy=self-supervised + supervised post-processing2021.05 | 2.79 | — | — | — | — | — | — | |
| OursSupervision Strategy=self-supervised (template, no labels)2021.05 | 3.31 | 3.51 | 3.28 | 3.5 | 3.03 | 2.97 | 3.55 | |
| Unsupervised Discovery of Object Landmarks (discovered landmarks)Supervision=Unsupervised2018.04 | 4.14 | 5.01 | 4.61 | 4.76 | 4.45 | 4.91 | 4.61 | |
| ZhangSupervision Strategy=self-supervised + supervised post-processing2021.05 | 4.14 | 5.01 | 4.61 | 4.76 | 4.45 | 4.91 | 4.61 | |
| Thewlis et al.Supervision=Unsupervised2018.04 | 7.51 | 7.54 | 8.56 | 7.26 | 6.47 | 7.93 | 5.4 | |
| ThewlisSupervision Strategy=self-supervised + supervised post-processing2021.05 | 7.51 | 7.54 | 8.56 | 7.26 | 6.47 | 7.93 | 5.4 |