Human Pose Estimation on COCO (minival)
77.5APCCM
Evaluation results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| CCMBackbone=HRNet-w48, Parameters=63.7M, GFLOPs=36.6, Input size=384 x 288, Model scale=Large2020.02 | 77.5 | 91.2 | 83.6 | 73 | 84 | 82.3 | |
| CCMBackbone=HRNet-w32, Parameters=28.6M, GFLOPs=7.92, Input size=256 x 192, Model scale=Medium2020.02 | 76.7 | 91.1 | 83.5 | 72.5 | 83 | 81.8 | |
| CCMBackbone=ResNet-152, Parameters=63.5M, GFLOPs=40.1, Input size=384 x 288, Model scale=Large2020.02 | 76.7 | 91.2 | 83.4 | 72.4 | 83.2 | 81.7 | |
| HRNetBackbone=HRNet-w48, Parameters=63.6M, GFLOPs=35.4, Input size=384 x 288, Model scale=Large2020.02 | 76.3 | 90.8 | 82.9 | 72.3 | 83.4 | 81.2 | |
| CCMBackbone=ResNeXt-50, Parameters=40.1M, GFLOPs=45.4, Input size=256 x 192, Model scale=Medium2020.02 | 74.5 | 90.4 | 81.2 | 70 | 81 | 79.6 | |
| HRNetBackbone=HRNet-w32, Parameters=28.5M, GFLOPs=7.68, Input size=256 x 192, Model scale=Medium2020.02 | 74.4 | 90.5 | 81.9 | 70.8 | 81 | 79.8 | |
| CCMBackbone=ResNet-50, Parameters=40.7M, GFLOPs=45.4, Input size=256 x 192, Model scale=Medium2020.02 | 74.3 | 90.3 | 81.3 | 70 | 80.6 | 79.6 | |
| Simple BaselineBackbone=ResNet-152, Parameters=68.6M, GFLOPs=35.9, Input size=384 x 288, Model scale=Large2020.02 | 74.3 | 89.6 | 81.1 | 70.5 | 79.7 | 79.7 | |
| Simple BaselineBackbone=ResNeXt-50, Parameters=33.5M, GFLOPs=8.35, Input size=256 x 192, Model scale=Medium2020.02 | 70.6 | 88.9 | 77.9 | 67.2 | 77.5 | 76.5 | |
| Simple BaselineBackbone=ResNet-50, Parameters=34.0M, GFLOPs=8.2, Input size=256 x 192, Model scale=Medium2020.02 | 70.4 | 88.6 | 78.3 | 67.1 | 77.2 | 76.3 | |
| CPNBackbone=ResNet-50, Parameters=27.0M, GFLOPs=6.2, Input size=256 x 192, Model scale=Medium2020.02 | 69.4 | — | — | — | — | — | |
| CCMBackbone=MobileNet-v2, Parameters=10.1M, GFLOPs=4.8, Input size=256 x 192, Model scale=Small2020.02 | 68.4 | 88 | 75.7 | 64.2 | 74.5 | 74.3 | |
| CCMBackbone=ShuffleNet-v2, Parameters=9.72M, GFLOPs=4.75, Input size=256 x 192, Model scale=Small2020.02 | 67.9 | 88.4 | 75.3 | 63.8 | 74 | 74.2 | |
| HourglassBackbone=8xHourglass, Parameters=25.1M, GFLOPs=14.3, Input size=256 x 192, Model scale=Medium2020.02 | 66.9 | — | — | — | — | — | |
| Simple BaselineBackbone=MobileNet-v2, Parameters=9.57M, GFLOPs=4.17, Input size=256 x 192, Model scale=Small2020.02 | 64.3 | 86.3 | 72.2 | 60.9 | 70.9 | 70.5 | |
| Simple BaselineBackbone=ShuffleNet-v2, Parameters=8.78M, GFLOPs=4.07, Input size=256 x 192, Model scale=Small2020.02 | 63.9 | 86.6 | 71.4 | 60.8 | 70.1 | 70.3 |