Research map / 3D vision / Human pose estimation
Equivariant Symmetry-Aware Head Pose Estimation for Fetal MRI
Ramya Muthukrishnan, Borjan Gagoski, Aryn Lee, Ellen Grant, Elfar Adalsteinsson, Benjamin Billot, Polina Golland
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| use E(3)-equivariant convolutions instead of standard convolutions (ablation) | E(3)-Pose · Navigators (trained on Research-Fetal) | rotation error (deg): 18.0 -> 9.4 | Improved |
| use E(3)-equivariant convolutions instead of standard convolutions (ablation) | E(3)-Pose · Navigators (trained on dHCP) | rotation error (deg): 80.2 -> 13.9 | Improved |
| use E(3)-equivariant convolutions instead of standard convolutions (ablation) | E(3)-Pose · Clinical-Young (trained on dHCP) | rotation error (deg): 53.3 -> 12.1 | Improved |
| model left-right axis as pseudovector instead of standard vector (ablation) | E(3)-Pose · Navigators (trained on Research-Fetal) | rotation error (deg): 10.9 -> 9.4 | Improved |
| model left-right axis as pseudovector instead of standard vector (ablation) | E(3)-Pose · Navigators (trained on dHCP) | rotation error (deg): 20.6 -> 13.9 | Improved |
| predict third basis direction e_x in addition to e_y and e_z (ablation) | E(3)-Pose · Navigators (trained on dHCP) | rotation error (deg): 26.1 -> 13.9 | Improved |
| predict third basis direction e_x in addition to e_y and e_z (ablation) | E(3)-Pose · Navigators (trained on Research-Fetal) | rotation error (deg): 11.9 -> 9.4 | Improved |
| use symmetry-aware pseudovector loss instead of vector loss term (ablation) | E(3)-Pose · Navigators (trained on Research-Fetal) | rotation error (deg): 12.1 -> 9.4 | Improved |
| use symmetry-aware pseudovector loss instead of vector loss term (ablation) | E(3)-Pose · Clinical-Young (trained on Research-Fetal) | rotation error (deg): 8.8 -> 9.1 | Hurt |
| use proposed loss instead of geodesic loss (ablation) | E(3)-Pose · Navigators (trained on dHCP) | rotation error (deg): 27.8 -> 13.9 | Improved |
| add simulated spin history artifact augmentation (ablation) | E(3)-Pose · Navigators (trained on Research-Fetal) | rotation error (deg): 16.2 -> 9.4 | Improved |
| add simulated spin history artifact augmentation (ablation) | E(3)-Pose · Navigators (trained on dHCP) | rotation error (deg): 31.9 -> 13.9 | Improved |
Closest papers by meaning
- CARL: A Framework for Equivariant Image Registration (CVPR 2025 · similarity 0.85)
- Uncertainty Estimation for Pretrained Medical Image Registration Models via Transformation Equivariance (NeurIPS 2026 · similarity 0.84)
- Augmented Equivariant Mesh Networks for Anatomical Segmentation (NeurIPS 2026 · similarity 0.83)
- Fully Convolutional Slice-to-Volume Reconstruction for Single-Stack MRI (CVPR 2024 · similarity 0.83)
- 3D Equivariant Pose Regression via Direct Wigner-D Harmonics Prediction (NeurIPS 2024 · similarity 0.83)
- On the Fourier analysis in the SO(3) space : the EquiLoPO Network (ICLR 2025 · similarity 0.82)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
