Human Pose Estimation on Penn-Action
98Head AccThin-Slicing Network S-infer
Evaluation results
| Method | Links | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|
| Thin-Slicing Network S-infertype=spatial inference2017.03 | 98 | 90.3 | 85.2 | 86.7 | 93.7 | 93.5 | 93.6 | — | — | 91.4 | |
| Thin-Slicing Network ST-infertype=spatio-temporal inference2017.03 | 98 | 97.3 | 95.1 | 94.7 | 97.1 | 97.1 | 96.9 | — | — | 96.5 | |
| Thin-Slicing Network baselinetype=pure ConvNet2017.03 | 97.9 | 94.9 | 76.8 | 72 | 95.9 | 88.8 | 85.1 | — | — | 87 | |
| Thin-Slicing Network ST-infer(+)type=baseline ConvNet after end-to-end training2017.03 | 97.9 | 91.1 | 91.3 | 90.9 | 92.5 | 94.4 | 94.5 | — | — | 92.8 | |
| Thin-Slicing Network ST-infer(*)type=independent training2017.03 | 97.9 | 89.7 | 84.4 | 86.5 | 93.4 | 93.7 | 93.8 | — | — | 91 | |
| Thin-Slicing Network ST-infer(2)temporal connection=across 2 frames2017.03 | 97.6 | 96.8 | 95.2 | 95.1 | 97 | 96.8 | 96.9 | — | — | 96.4 | |
| Gkioxari et al.2016.03 | 95.6 | 93.8 | 90.4 | 90.7 | 91.8 | 90.8 | 91.5 | 91.8 | — | — | |
| Nie et al.2017.03 | 95.6 | 93.8 | 90.4 | 90.7 | 91.8 | 90.8 | 91.5 | — | — | 91.8 | |
| ACPSBinary=Cond(5) + AS, Unary=Cond(5), Features=CCF2016.03 | 89.1 | 86.4 | 73.9 | 73 | 85.3 | 79.9 | 80.3 | 81.1 | 64.8 | — | |
| Gkioxari et al.2017.03 | 89.1 | 86.4 | 73.9 | 73 | 85.3 | 79.9 | 80.3 | — | — | 81.1 | |
| ACPSBinary=Indep., Unary=Indep., Features=CCF2016.03 | 84.5 | 81.3 | 66.2 | 62.6 | 82.4 | 75.1 | 76.5 | 75.5 | 57.3 | — | |
| Nie et al.2016.03 | 64.2 | 55.4 | 33.8 | 24.4 | 56.4 | 54.1 | 48 | 48 | — | — | |
| Yang and Ramanan2017.03 | 64.2 | 55.4 | 33.8 | 24.4 | 56.4 | 54.1 | 48 | — | — | 48 | |
| Park & Ramanan2016.03 | 62.8 | 52 | 32.3 | 23.3 | 53.3 | 50.2 | 43 | 45.3 | — | — | |
| Park and Ramanan2017.03 | 62.8 | 52 | 32.3 | 23.3 | 53.3 | 50.2 | 43 | — | — | 45.3 | |
| Yang & Ramanan2016.03 | 57.9 | 51.3 | 30.1 | 21.4 | 52.6 | 49.7 | 46.2 | 44.2 | — | — |