Research map / Label-efficient and robust learning / Recognition and long-tail
Part-Aligned Bilinear Representations for Person Re-Identification
Yumin Suh, Jingdong Wang, Siyu Tang, Tao Mei, Kyoung Mu Lee
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 maps via bilinear pooling to appearance-only baseline (ablation) | GoogLeNet appearance baseline · Market-1501 | rank-1 accuracy: 81.6 -> 90.2 | Improved |
| add part maps via bilinear pooling to appearance-only baseline (ablation) | GoogLeNet appearance baseline · MARS | mAP: 63.1 -> 72.2 | Improved |
| add part maps via bilinear pooling to appearance-only baseline (ablation) | GoogLeNet appearance baseline · DukeMTMC | mAP: 50.6 -> 64.2 | Improved |
| fine-tune the pose sub-network instead of fixing it (ablation) | part map extractor · Market-1501, MARS, Duke | accuracy | Improved |
| replace concat+averagepool+linear with bilinear pooling (ablation) | aggregation of appearance and part maps · Market-1501 | accuracy | Improved |
| use joint+limb+internal maps instead of joints only (ablation) | OpenPose pose sub-network · Market-1501 (pose sub-network) | mAP: 75.6 -> 76.0 | Improved |
| use joint+limb+internal maps instead of internal only (ablation) | OpenPose pose sub-network · Market-1501 (pose sub-network) | mAP: 74.3 -> 76.0 | Improved |
| use limb maps only instead of full maps (ablation) | OpenPose pose sub-network · Market-1501 (pose sub-network) | rank-1 accuracy: 90.2 -> 90.5 | No clear effect |
| add dilation filters to appearance extractor (ablation) | Inception-V1 + OpenPose · Market-1501 (single query) | mAP: 76.0 -> 79.6 | Improved |
| add dilation filters to appearance extractor (ablation) | Inception-V1 + OpenPose · DukeMTMC | mAP: 64.2 -> 69.3 | Improved |
| add dilation filters to appearance extractor (ablation) | Inception-V1 + OpenPose · MARS | mAP: 72.2 -> 75.9 | Improved |
| add re-ranking | Inception-V1 + OpenPose + dilation · Market-1501 (single query) | mAP: 79.6 -> 89.9 | Improved |
Closest papers by meaning
- Deeply-Learned Part-Aligned Representations for Person Re-Identification (ICCV 2017 · similarity 0.92)
- Multi-Task Learning With Coarse Priors for Robust Part-Aware Person Re-Identification (TPAMI 2020 · similarity 0.91)
- Semantics-Aligned Representation Learning for Person Re-Identification (AAAI 2020 · similarity 0.90)
- Similarity Learning With Spatial Constraints for Person Re-Identification (CVPR 2016 · similarity 0.90)
- Densely Semantically Aligned Person Re-Identification (CVPR 2019 · similarity 0.89)
- SEAS: ShapE-Aligned Supervision for Person Re-Identification (CVPR 2024 · similarity 0.87)
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.
