Research map / 3D vision / Human pose estimation
Chirality Nets for Human Pose Regression
Raymond Yeh, Yuan-Ting Hu, Alexander Schwing
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 |
|---|---|---|---|
| Replace the temporal convolutional network layers with chirality-equivariant layers | Human3.6M 2D-to-3D pose estimation · Human3.6M, Protocol 1 (MPJPE) | MPJPE: 0.1 mm | Improved |
| Replace the baseline with the chirality-equivariant single-frame model | Human3.6M 2D-to-3D pose estimation · Human3.6M, single-frame comparison | MPJPE: 0.4 mm | Improved |
| Use chirality-equivariant layers instead of the baseline without test-time augmentation | Human3.6M 2D-to-3D pose estimation · Human3.6M, Protocol 1 | MPJPE: 1 mm | Improved |
| Use chirality-equivariant temporal convolution | HumanEva-I 2D-to-3D pose estimation · HumanEva-I, Protocol 1 | MPJPE: 0.8mm | Improved |
| Use chirality-equivariant temporal convolution | HumanEva-I 2D-to-3D pose estimation · HumanEva-I, Protocol 2 | P-MPJPE: 0.2 mm | Improved |
| Use chirality-equivariant layers | Human3.6M 2D-to-3D pose estimation · Human3.6M, Protocol 2 | P-MPJPE: 0.3mm | Hurt |
| Use chirality-equivariant layers | Human3.6M limited-data pose estimation · Human3.6M, training data from S1 at 0.1% | MPJPE | Hurt |
| Use chirality-equivariant layers with batch-size 32 | Human3.6M limited-data pose estimation · Human3.6M, S1 0.1% | MPJPE: 100.4mm; 102.3mm | Improved |
| Replace the non-chiral sequence model with the chirality-equivariant model | Penn Action pose forecasting · Penn Action | PCK@0.05 | Improved |
| Add train and test-time augmentation to the non-chiral forecasting baseline | Penn Action pose forecasting · Penn Action | PCK@0.05 | Improved |
| Add a chiral-invariance layer and chirality-equivariant layers before it | Ours-Conv skeleton action recognition · Kinetics-400 | Top-1 accuracy | Improved |
| Add a chiral-invariance layer and chirality-equivariant layers before it | Ours-Conv skeleton action recognition · Kinetics-400 | Top-5 accuracy | Improved |
Closest papers by meaning
- CubeNet: Equivariance to 3D Rotation and Translation (ECCV 2018 · similarity 0.86)
- Semantic Graph Convolutional Networks for 3D Human Pose Regression (CVPR 2019 · similarity 0.86)
- 3D Human Pose Estimation From a Single Image via Distance Matrix Regression (CVPR 2017 · similarity 0.86)
- Towards Symmetry-sensitive Pose Estimation: A Rotation Representation for Symmetric Object Classes (IJCV 2026 · similarity 0.86)
- Chirality in Action: Time-Aware Video Representation Learning by Latent Straightening (NeurIPS 2025 · similarity 0.85)
- Neural Isometries: Taming Transformations for Equivariant ML (NeurIPS 2024 · similarity 0.84)
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.
