Research map / 3D vision / Human pose estimation
Human Pose Estimation With Parsing Induced Learner
Xuecheng Nie, Jiashi Feng, Yiming Zuo, Shuicheng Yan
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 PIL to the VGG16 pose model (ablation) | VGG16 human pose estimation · LIP validation set | PCK: 69.1 -> 75.0 | Improved |
| Fuse parsing features by addition, multiplication, or concatenation instead of PIL (ablation) | VGG16 human pose estimation · LIP validation set | PCK: 69.1 -> 68.6; 69.1 -> 63.7; 69.1 -> 66.1 | Hurt |
| Use multi-task learning instead of PIL (ablation) | VGG16 human pose estimation · LIP validation set | PCK: 69.1 -> 65.3 | Hurt |
| Add a residual module without PIL (ablation) | VGG16 human pose estimation · LIP validation set | PCK: 69.1 -> 69.8 | Improved |
| Add PIL (ablation) | Hourglass pose models · LIP validation set | PCK: 78.8 -> 82.2; 80.8 -> 83.0; 84.5 -> 85.3; 84.9 -> 85.6 | Improved |
| Add PIL with a smaller parsing network rather than a larger one (ablation) | Hourglass pose models · LIP validation set | PCK: HG-1s-1u: 84.5 -> 85.0; HG-1s-2u: 84.5 -> 85.3; HG-1s-4u: 84.5 -> 85.4; HG-1s-8u: 84.5 -> 85.5 | Improved |
| Improve parsing-network depth and parsing quality (ablation) | HG-4s-2u pose model with PIL · LIP validation set | PCK: 41.1 -> 43.9 mIOU; 85.0 -> 85.5 PCK | Improved |
| Add PIL (ablation) | Hourglass multi-person pose model · extended PASCAL-Person-Part | mAP: 39.6 -> 41.0 | Improved |
| Transfer PIL pretrained on LIP to MPII without fine-tuning (ablation) | HG-8s-1u pose model · MPII validation set | PCKh: 90.2 -> 91.0 | Improved |
| Transfer PIL pretrained on LIP to MPII (ablation) | human pose estimation model · MPII training set | training accuracy and loss | Improved |
| Use the LIP-trained PIL on MPII (ablation) | human pose estimation model · MPII testing set | PCKh: 92.0 -> 92.4 | Mixed |
Closest papers by meaning
- Look into Person: Joint Body Parsing & Pose Estimation Network and a New Benchmark (TPAMI 2018 · similarity 0.88)
- Pose-Guided Human Parsing by an AND/OR Graph Using Pose-Context Features (AAAI 2016 · similarity 0.87)
- Mutual Learning to Adapt for Joint Human Parsing and Pose Estimation (ECCV 2018 · similarity 0.87)
- Look Into Person: Self-Supervised Structure-Sensitive Learning and a New Benchmark for Human Parsing (CVPR 2017 · similarity 0.87)
- Neural Architecture Search for Joint Human Parsing and Pose Estimation (ICCV 2021 · similarity 0.86)
- Fast Adaptation for Human Pose Estimation via Meta-Optimization (CVPR 2024 · 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.
