2D Human Pose Estimation on MRT
75.6APViTPose (LSP + STP)
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| ViTPose (LSP + STP)Model size=Base, Fine-tuning data=LSP + STP, N images=12,5202026.04 | 75.6 | 91.2 | 81.6 | 94.5 | |
| ViTPose (LSP + STP)Model size=Small, Fine-tuning data=LSP + STP, N images=12,5202026.04 | 73.1 | 90.2 | 79.1 | 93.5 | |
| ViTPose (RePoGen)Model size=Small, Fine-tuning data=RePoGen, N images=3,0002026.04 | 71.7 | 87.4 | 78.3 | 91.1 | |
| ViTPose (LSP)Model size=Small, Fine-tuning data=LSP, N images=10,0002026.04 | 66 | 87.7 | 73.7 | 92.5 | |
| ViTPose (Baseline)Model size=Base, Fine-tuning data=None, N images=-2026.04 | 65.6 | 87.8 | 71.9 | 91.8 | |
| ViTPose (STP)Model size=Small, Fine-tuning data=STP, N images=2,5202026.04 | 60.1 | 84.2 | 67.6 | 88.7 | |
| ViTPose (Baseline)Model size=Small, Fine-tuning data=None, N images=-2026.04 | 55.8 | 80.5 | 63.5 | 85.3 |