Research map / 3D vision / Human pose estimation
Keypoint Communities
Duncan Zauss, Sven Kreiss, Alexandre Alahi
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 local harmonic centrality keypoint weighting (ablation) | OpenPifPaf ShuffleNetV2k16 · COCO WholeBody val | AP (%): 53.6 -> 55.3 | Improved |
| compute centrality on the ego graph of radius three rather than the global graph (ablation) | keypoint weighting · COCO WholeBody val | AP gain vs baseline (%): -1.2% | Improved |
| use centrality-derived weights instead of hand-crafted weights (ablation) | keypoint weighting · COCO WholeBody val | AP gain vs baseline (%): -1.2% | Improved |
| add keypoint weighting (vs equal weights) on foot keypoints (ablation) | OpenPifPaf · COCO WholeBody val foot | AP gain vs baseline (%): -8.1% | Improved |
| add keypoint weighting (vs equal weights) on face keypoints (ablation) | OpenPifPaf · COCO WholeBody val face | AP gain vs baseline (%): +1.5% | Hurt |
| add keypoint weighting (vs equal weights) on body keypoints (ablation) | OpenPifPaf · COCO WholeBody val body | AP gain vs baseline (%): -4.8% | Improved |
| use global graph weights instead of ego graph on hand keypoints (ablation) | keypoint weighting · COCO WholeBody val hand | AP gain vs baseline (%): +2.6% | Hurt |
| lower seed threshold from 0.5 to 0.2 in decoding (ablation) | OpenPifPaf decoder · COCO WholeBody val | AP (%): 58.4 -> 60.4 | Improved |
| raise CAF threshold from 0.001 to 0.01 at seed threshold 0.5 (ablation) | OpenPifPaf decoder · COCO WholeBody val | AP (%): 58.4 -> 54.6 | Hurt |
| use bottom-up single-network model instead of top-down ZoomNet | Ours vs ZoomNet · COCO WholeBody val | AP (%): 54.1 -> 60.4 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- OpenPose — OpenPose: Realtime Multi-Person 2D Pose Estimation Using Part Affinity Fields (TPAMI 2019) · better on 5 of 5 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Simple Pose: Rethinking and Improving a Bottom-up Approach for Multi-Person Pose Estimation (AAAI 2020 · similarity 0.91)
- Rethinking Keypoint Representations: Modeling Keypoints and Poses as Objects for Multi-Person Human Pose Estimation (ECCV 2022 · similarity 0.89)
- Group Pose: A Simple Baseline for End-to-End Multi-Person Pose Estimation (ICCV 2023 · similarity 0.88)
- Towards Accurate Multi-Person Pose Estimation in the Wild (CVPR 2017 · similarity 0.88)
- ProbPose: A Probabilistic Approach to 2D Human Pose Estimation (CVPR 2025 · similarity 0.88)
- KeyPose: Category-Level 6D Object Pose Estimation with Self-Adaptive Keypoints (AAAI 2025 · 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.
