Research map / 3D vision / Face and shape
Learning Articulated Shape With Keypoint Pseudo-Labels From Web Images
Anastasis Stathopoulos, Georgios Pavlakos, Ligong Han, Dimitris N. Metaxas
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 consistency-selected web keypoint pseudo-labels to ACSM-ours | ACSM-ours · Pascal Horse, Cow, and Sheep | AUC: 50.8 -> 55.9; 47.6 -> 52.6; 46.8 -> 53.0 | Improved |
| Add consistency-selected web keypoint pseudo-labels to ACSM-ours | ACSM-ours · Animal Pose Horse, Cow, and Sheep | AUC: 66.7 -> 75.1; 59.6 -> 69.6; 57.2 -> 60.2 | Improved |
| Add consistency-selected web keypoint pseudo-labels to ACSM-ours | ACSM-ours · COCO Giraffe and Bear | AUC: 79.1 -> 86.6; 59.2 -> 63.4 | Improved |
| Use all web pseudo-labels rather than curating a subset (ablation) | ACSM-ours · Pascal Horse, Cow, and Sheep | AUC: 50.8 -> 50.2; 47.6 -> 43.9; 46.8 -> 43.5 | Hurt |
| Use all web pseudo-labels rather than curating a subset (ablation) | ACSM-ours · Animal Pose Sheep | AUC: 57.2 -> 29.1 | Hurt |
| Use all web pseudo-labels rather than curating a subset (ablation) | ACSM-ours · COCO Giraffe and Bear | AUC: 79.1 -> 75.6; 59.2 -> 58.2 | Hurt |
| Select pseudo-labels by consistency rather than keypoint confidence alone (ablation) | ACSM-ours with web pseudo-labels · Pascal, Animal Pose, and COCO | AUC | Improved |
| Use keypoint pseudo-labels from unlabeled images | CMR · CUB | 3D reconstruction performance | Improved |
| Reduce the labeled set while retaining 3K selected pseudo-labeled images (ablation) | ACSM horse reconstruction · Pascal Horse | AUC | Improved |
| Use CF-CM² consistency filtering (ablation) | ACSM-ours with selected web pseudo-labels · Pascal Horse camera viewpoint | err_R: 30.2 -> 28.1 | Improved |
| Use CF-CM² consistency filtering (ablation) | ACSM-ours with selected web pseudo-labels · Pascal Cow camera viewpoint | err_R: 34.7 -> 32.7 | Improved |
Closest papers by meaning
- DensePose 3D: Lifting Canonical Surface Maps of Articulated Objects to the Third Dimension (ICCV 2021 · similarity 0.90)
- MagicPony: Learning Articulated 3D Animals in the Wild (CVPR 2023 · similarity 0.89)
- Unsupervised 3D Structure Inference from Category-Specific Image Collections (CVPR 2024 · similarity 0.88)
- ARTIC3D: Learning Robust Articulated 3D Shapes from Noisy Web Image Collections (NeurIPS 2023 · similarity 0.88)
- Ponymation: Learning Articulated 3D Animal Motions from Unlabeled Online Videos (ECCV 2024 · similarity 0.88)
- SAOR: Single-View Articulated Object Reconstruction (CVPR 2024 · similarity 0.87)
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.
