Research map / Detection and segmentation / Open-set and few-shot detection
InclusiveHuman-10K: Towards Inclusive Human Parsing Beyond the Intact-Limb Assumption
Heming Du, Jiaying Ying, Xiaofeng Cao, Zhu Li, Yuanyuan Liu, Kaihao Zhang, Xin Chen, 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 |
|---|---|---|---|
| fine-tune on IH-10K with Common vs Disability-Related split (gap) | UniParser R101 · IH-10K | mIoU: 58.10 -> 28.19 | Hurt |
| scale backbone ResNet50 to ResNet101 | UniParser · IH-10K Disability-Related | mIoU: 25.18 -> 28.19 | Improved |
| scale backbone ResNet50 to ResNet101 | AIParsing · IH-10K Disability-Related | mIoU: 17.51 -> 15.60 | Hurt |
| scale backbone ResNet50 to ResNet101 | AIParsing · IH-10K All | mIoU: 30.97 -> 31.49 | Improved |
| scale backbone size | Mask2Former/SegFormer/SegNeXt · IH-10K Disability-Related | mIoU | Mixed |
| fine-tune SAM3 on IH-10K | SAM3 · IH-10K All | mIoU: 19.55 -> 33.38 | Improved |
| fine-tune SAM3 on IH-10K | SAM3 · IH-10K Disability-Related | mIoU: 5.05 -> 14.20 | Improved |
| tune Grounded-SAM2 on IH-10K | Grounded-SAM2 · IH-10K All | mIoU: 4.66 -> 6.68 | Improved |
| use ground-truth boxes with SAM2 instead of point prompts | SAM2 Tiny · IH-10K All | mIoU: 27.09 -> 77.06 | Improved |
| add 10% noise to box prompts (ablation) | SAM2 · IH-10K | mIoU: 75.32 -> 73.17 | No clear effect |
| use zero-shot Grounded-SAM2 boxes instead of ground-truth boxes (ablation) | SAM2 · IH-10K | mIoU: 75.32 -> 4.94 | Hurt |
| increase Grounded SAM backbone from Base to Huge | Grounded SAM · IH-10K | mIoU: 5.99 -> 6.10 | No clear effect |
Closest papers by meaning
- IVQA-LD: Inclusive Multimodal Understanding for Population with Limb Deficiency (ICML 2026 · similarity 0.90)
- InclusiveVidPose: Bridging the Pose Estimation Gap for Individuals with Limb Deficiencies in Video-Based Motion (ICLR 2026 · similarity 0.89)
- LDPose: Towards Inclusive Human Pose Estimation for Limb-Deficient Individuals in the Wild (ICCV 2025 · similarity 0.89)
- HumanRefiner: Benchmarking Abnormal Human Generation and Refining with Coarse-to-fine Pose-Reversible Guidance (ECCV 2024 · similarity 0.87)
- Same Person, Different Depiction: Counterfactual Evaluation of Vision-Language Models on Individuals with Limb Deficiencies (ECCV 2026 · similarity 0.87)
- SapiensID 2.0: Aligning Human Recognition Foundation Models with Human Perception (NeurIPS 2026 · 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.
