Research map / 3D vision / Human pose estimation
SelfGeo: Self-supervised and Geodesic-consistent Estimation of Keypoints on Deformable Shapes
Mohammad Zohaib, Luca Cosmo, Alessio Del Bue
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 |
|---|---|---|---|
| add surface loss to the shape loss (ablation) | SelfGeo · mini-CAPE | inclusivity / coverage / T_con: removal: -8.66 / -6.45 / -5.01 | Improved |
| add reconstruction loss (ablation) | SelfGeo · mini-CAPE | reconstruction error: removal: +0.230 | Improved |
| add temporal smoothing loss (ablation) | SelfGeo · mini-CAPE | T_con: removal: -2.11 | Improved |
| add geodesic distance loss (ablation) | SelfGeo · mini-CAPE | inclusivity / coverage: removal: -3.74 / -4.14 | Improved |
| add full deformation loss (geodesic plus smoothing) (ablation) | SelfGeo · mini-CAPE | coverage / T_con: removal: -5.94 / -3.82 | Improved |
| add coverage loss and shape loss (ablation) | SelfGeo · mini-CAPE | training success | Improved |
| use PointNet++ instead of PointNet backbone (ablation) | SelfGeo · mini-CAPE | inclusivity / coverage / T_con: PointNet: -1.31 / -0.53 / -1.95 | Improved |
| increase Gaussian noise variance from 0.01 to 0.05 (ablation) | SelfGeo · CAPE noisy PCDs | inclusivity: 81.20 -> 54.07 | Hurt |
| downsample point cloud by FPS up to x32 (ablation) | SelfGeo · CAPE downsampled PCDs | inclusivity: 82.67 -> 62.13 | Hurt |
| SelfGeo versus SC3K and ULCS on real ITOP data | unsupervised keypoint estimation · ITOP | T_con: 42.86 -> 80.64 (vs SC3K) | Improved |
Closest papers by meaning
- KeypointDeformer: Unsupervised 3D Keypoint Discovery for Shape Control (CVPR 2021 · similarity 0.90)
- SC3K: Self-supervised and Coherent 3D Keypoints Estimation from Rotated, Noisy, and Decimated Point Cloud Data (ICCV 2023 · similarity 0.88)
- Key-Grid: Unsupervised 3D Keypoints Detection using Grid Heatmap Features (NeurIPS 2024 · similarity 0.87)
- UKPGAN: A General Self-Supervised Keypoint Detector (CVPR 2022 · similarity 0.87)
- KeypointDETR: An End-to-End 3D Keypoint Detector (ECCV 2024 · similarity 0.86)
- SPU-PMD: Self-Supervised Point Cloud Upsampling via Progressive Mesh Deformation (CVPR 2024 · similarity 0.86)
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.
