Research map / 3D vision / Human pose estimation
Robust Pose Estimation in Crowded Scenes with Direct Pose-Level Inference
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 Pose Refinement (PR) module (ablation) | PINet (PPG) · OCHuman test | AP: 53.1 -> 58.8 | Helped |
| add Pose Fusion (PF) (ablation) | PINet (PPG+PR) · OCHuman test | AP: 58.8 -> 59.8 | Helped |
| add Part-based Pose Generation (3 parts) vs center-point regression (ablation) | PINet · OCHuman test | AP: 47.6 -> 53.1 | Helped |
| use part-large (3 parts) instead of head-only partition (ablation) | PINet PPG · OCHuman test | AP: 49.8 -> 53.1 | Helped |
| use finer partition (part-small, 6 parts) (ablation) | PINet PPG · OCHuman test | AP: 53.1 -> 50.3 | Hurt |
| use finest partition (13 keypoint parts) (ablation) | PINet PPG · OCHuman test | AP: 53.1 -> 40.8 | Hurt |
| add static skeleton graph to PR (ablation) | PINet PR · OCHuman test | AP: 56.8 -> 57.7 | Helped |
| make graph dynamic instead of static (ablation) | PINet PR · OCHuman test | AP: 57.7 -> 59.8 | Helped |
| apply PF to SPM (ablation) | SPM · OCHuman test | AP: 47.6 -> 48.1 | Helped |
| apply PF to DEKR (ablation) | DEKR · OCHuman test | AP: 52.2 -> 53.0 | Helped |
| apply PF to PINet without PR (ablation) | PINet (w/o PR) · OCHuman test | AP: 53.1 -> 55.8 | Helped |
| PINet vs DEKR on COCO val (non-crowded) | PINet · COCO val | AP: 68.0 -> 67.4 | No clear effect |
Closest papers by meaning
- CrowdPose: Efficient Crowded Scenes Pose Estimation and a New Benchmark (CVPR 2019 · similarity 0.88)
- RSGNet: Relation based Skeleton Graph Network for Crowded Scenes Pose Estimation (AAAI 2021 · similarity 0.88)
- Simple Pose: Rethinking and Improving a Bottom-up Approach for Multi-Person Pose Estimation (AAAI 2020 · similarity 0.87)
- DeepCut: Joint Subset Partition and Labeling for Multi Person Pose Estimation (CVPR 2016 · similarity 0.87)
- Generating Multiple Hypotheses for 3D Human Pose Estimation With Mixture Density Network (CVPR 2019 · similarity 0.86)
- ArtTrack: Articulated Multi-Person Tracking in the Wild (CVPR 2017 · 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.