Metric: MAE (trained with BIWI data) (lower is better)
| # | Model↕ | MAE (trained with BIWI data)▲ | Extra Data | Paper | Date↕ | Code |
|---|---|---|---|---|---|---|
| 1 | VGGNet-RNN | 1.413 | Yes | - | - | - |
| 2 | Direct Regression | 2.54 | No | Deep Ordinal Regression with Label Diversity | 2020-06-29 | Code |
| 3 | 6DRepNet | 2.66 | No | 6D Rotation Representation For Unconstrained Hea... | 2022-02-25 | Code |
| 4 | DDD-Pose | 2.8 | No | - | - | - |
| 5 | Hybrid Coarse-Fine | 3.0174 | No | Hybrid coarse-fine classification for head pose ... | 2019-01-21 | Code |
| 6 | EHPNet | 3.43 | No | An Effective Deep Network for Head Pose Estimati... | 2022-10-25 | - |
| 7 | DAD-3DNet | 3.98 | No | DAD-3DHeads: A Large-scale Dense, Accurate and D... | 2022-04-07 | Code |
| 8 | 2DCube | 4.31 | No | An Intuitive and Unconstrained 2D Cube Represent... | 2022-12-07 | - |
| 9 | Multi-Loss ResNet50 | 4.895 | No | Fine-Grained Head Pose Estimation Without Keypoi... | 2017-10-02 | Code |