Table 1 Comparison of different models on MPII human and OCHuman datasets for pose estimation Task.
From: Deep learning-based approaches for human pose estimation in interdisciplinary physics applications
Model | MPII Human Dataset | OCHuman Dataset | ||||||
|---|---|---|---|---|---|---|---|---|
Accuracy | Recall | F1 Score | AUC | Accuracy | Recall | F1 Score | AUC | |
OpenPose16 | 84.21 | 82.13 | 83.67 | 85.89 | 80.32 | 77.89 | 79.23 | 81.45 |
AlphaPose51 | 86.34 | 84.56 | 85.21 | 88.12 | 82.54 | 80.11 | 81.67 | 84.23 |
DeepCut52 | 83.78 | 81.42 | 82.94 | 85.03 | 79.12 | 76.23 | 77.56 | 80.14 |
HRNet18 | 88.41 | 86.35 | 87.12 | 89.78 | 83.67 | 81.42 | 82.83 | 85.94 |
SimpleBaseline53 | 85.67 | 83.23 | 84.15 | 87.34 | 81.56 | 78.91 | 80.23 | 83.45 |
DARKPose54 | 87.12 | 85.34 | 86.03 | 88.45 | 82.89 | 80.57 | 81.94 | 84.78 |
ViTPose55 | 89.34 | 87.23 | 87.89 | 90.12 | 84.45 | 82.67 | 83.34 | 86.01 |
TokenPose56 | 89.78 | 87.89 | 88.01 | 90.34 | 84.89 | 83.12 | 84.02 | 86.45 |
MixSTE57 | 90.01 | 88.12 | 88.45 | 90.67 | 85.12 | 83.78 | 84.73 | 87.02 |
Ours (HSTPN) | 90.21 | 88.45 | 89.12 | 91.56 | 85.67 | 83.78 | 84.92 | 87.34 |
