Research map / 3D vision / Human pose estimation
InclusiveVidPose: Bridging the Pose Estimation Gap for Individuals with Limb Deficiencies in Video-Based Motion
Heming Du, Jiaying Ying, Sen Wang, Xue Li, Kaihao Zhang, Xin Yu
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 COCO pretraining initialization (ablation) | ViTPose-H · InclusiveVidPose 8 residual endpoints | AP/AR: 82.4/85.2 to 84.2/86.7 | Improved |
| add COCO pretraining initialization (ablation) | ViTPose-H · InclusiveVidPose 17 standard joints | AP/AR: 85.2/86.6 to 86.7/87.9 | Improved |
| add COCO pretraining initialization (ablation) | ViTPose-H · InclusiveVidPose 25 keypoints | AP/AR: 84.8/86.3 to 86.3/87.6 | Improved |
| add COCO to training data (joint training) (ablation) | ViTPose-H · InclusiveVidPose 17 standard joints | AP/AR: AP 86.7 to 87.0, AR 87.9 to 88.3 | Improved |
| add COCO to training data (joint training) (ablation) | ViTPose-H · InclusiveVidPose 8 residual endpoints | AP/AR: AP 84.2 to 82.2, AR 86.7 to 85.7 | Hurt |
| add COCO to training data (joint training) (ablation) | ViTPose-H · InclusiveVidPose 25 keypoints | AP: 86.3 to 86.5 | No clear effect |
| train extra residual-limb keypoints in head instead of 17 only | pose models · COCO | accuracy | Hurt |
| use video-based multi-frame model DSTA instead of DCPose | DSTA · InclusiveVidPose video | mean AP: 43.2 to 43.7 | No clear effect |
| use larger backbone ViT-H instead of ViT-S | ViTPose · InclusiveVidPose | LiCC: 67.1 to 73.6 on LiCC | Mixed |
Closest papers by meaning
- LDPose: Towards Inclusive Human Pose Estimation for Limb-Deficient Individuals in the Wild (ICCV 2025 · similarity 0.94)
- InclusiveHuman-10K: Towards Inclusive Human Parsing Beyond the Intact-Limb Assumption (ECCV 2026 · similarity 0.89)
- Prosthesis-Aware 3D Human Pose Estimation: A Dataset and Benchmark for RSP Users (ECCV 2026 · similarity 0.89)
- ResiHMR: Residual-Limb Aware Single-Image 3D Human Mesh Recovery for Individuals with Limb Loss (CVPR 2026 · similarity 0.87)
- HiPoser: 3D Human Pose Estimation with Hierarchical Shared Learning at Parts-Level Using Inertial Measurement Units (AAAI 2025 · similarity 0.85)
- FlexPose: Pose Distribution Adaptation with Limited Guidance (AAAI 2025 · similarity 0.85)
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.
