3D Human Pose Estimation on New Radar Dataset P1 (test)
5.027MPJPE (0°)RPP-Gauss.-Gauss.-Cov.
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| RPP-Gauss.-Gauss.-Cov.Latent Prior=Gaussian, Log-likelihood=Gaussian, full covariance output=true2025.08 | 5.027 | 6.264 | 7.22 | 6.17 | |
| RPP-Gauss.-Gauss.Latent Prior=Gaussian, Log-likelihood=Gaussian2025.08 | 5.127 | 6.374 | 7.022 | 6.174 | |
| RPP-Laplace-Gauss.Latent Prior=Laplace, Log-likelihood=Gaussian2025.08 | 5.203 | 6.583 | 7.166 | 6.317 | |
| Ho et al.2025.08 | 5.946 | 7.438 | 7.945 | 7.11 | |
| Engel et al.Pointcloud-based approach=true2025.08 | 6.643 | 7.662 | 9.695 | 8.031 | |
| RPP-Gauss.-LaplaceLatent Prior=Gaussian, Log-likelihood=Laplace2025.08 | 7.014 | 7.32 | 8.126 | 7.487 | |
| RPP-Normalizing-FlowsLatent Prior=Normalizing-Flows2025.08 | 7.097 | 7.023 | 7.196 | 7.072 | |
| RPP-Laplace-LaplaceLatent Prior=Laplace, Log-likelihood=Laplace2025.08 | 7.783 | 7.098 | 7.321 | 7.401 | |
| Evidential Regression2025.08 | 11.285 | 11.086 | 11.396 | 11.256 |