Research map / Detection and segmentation / CNN features
Pose Proposal Networks
Taiki Sekii
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 part-scale information via box IoU p(I|R,k,i) and box-based NMS (ablation) | PPN ResNet-18 · MPII Multi-Person validation | mAP: 72.4 -> 75.5 | Improved |
| add part-scale information via box IoU p(I|R,k,i) and box-based NMS (ablation) | PPN ResNet-50 · MPII Multi-Person validation | mAP: 75.9 -> 78.1 | Improved |
| add part-scale information via box IoU p(I|R,k,i) and box-based NMS (ablation) | PPN ResNet-101 · MPII Multi-Person validation | mAP: 77.5 -> 78.7 | Improved |
| add global-context parts association (route-based score) instead of local-only (ablation) | PPN ResNet-18 · MPII Multi-Person validation | mAP: 73.3 -> 75.5 | Improved |
| add global-context parts association (route-based score) instead of local-only (ablation) | PPN ResNet-50 · MPII Multi-Person validation | mAP: 77.5 -> 78.1 | Improved |
| add global-context parts association (route-based score) instead of local-only (ablation) | PPN ResNet-101 · MPII Multi-Person validation | mAP: 78.4 -> 78.7 | Improved |
| add global-context parsing, compared across depth (ablation) | PPN ResNet-18 vs ResNet-101 · MPII Multi-Person validation | mAP: -2.2% vs -0.3% | Improved |
| deepen backbone from ResNet-18 to ResNet-50 to ResNet-101 | PPN · MPII Multi-Person test (entire) | mAP: 72.8 -> 75.9 -> 76.6 | Improved |
| deepen backbone from ResNet-18 to ResNet-50 to ResNet-101 | PPN · MPII Multi-Person test (288-image subset) | mAP: 76.6 -> 79.4 -> 79.1 | Mixed |
| use ResNet-18 PPN instead of ResNet-152 DeeperCut part detector | PPN · MPII Single-Person test | PCKh: 88.1 vs 88.5 PCKh; 388 vs 34 FPS | Mixed |
Closest papers by meaning
- Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks (TPAMI 2016 · similarity 0.87)
- Human Pose Estimation via Convolutional Part Heatmap Regression (ECCV 2016 · similarity 0.87)
- CAP-Net: A Unified Network for 6D Pose and Size Estimation of Categorical Articulated Parts from a Single RGB-D Image (CVPR 2025 · similarity 0.87)
- Convolutional Pose Machines (CVPR 2016 · similarity 0.86)
- Single-Network Whole-Body Pose Estimation (ICCV 2019 · similarity 0.86)
- Detection Based Part-level Articulated Object Reconstruction from Single RGBD Image (NeurIPS 2023 · 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.
