Research map / Label-efficient and robust learning / Recognition and long-tail
Deep Attributes Driven Multi-camera Person Re-identification
Chi Su, Shiliang Zhang, Junliang Xing, Wen Gao, Qi Tian
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 attributes triplet loss fine-tuning and combined-set stage (ablation) | Stage 1 dCNN · VIPeR | rank-1 accuracy: 34.5 -> 37.9 | Improved |
| add attributes triplet loss fine-tuning and combined-set stage (ablation) | Stage 1 dCNN · PRID | rank-1 accuracy: 18.7 -> 20.1 | Improved |
| add attributes triplet loss fine-tuning and combined-set stage (ablation) | Stage 1 dCNN · GRID | rank-1 accuracy: 16.9 -> 19.1 | Improved |
| add the three-stage SSDAL training (ablation) | Stage 1 dCNN · VIPeR, PRID, GRID | attribute classification accuracy: 1.2% | Improved |
| add stage-3 combined training with self-labeled data only (no stage 2) (ablation) | Stage 1 dCNN · VIPeR, PRID, GRID | attribute classification accuracy | No clear effect |
| add stage-2 triplet fine-tuning without stage 3 (ablation) | Stage 1 dCNN · VIPeR, PRID, GRID | attribute classification accuracy | No clear effect |
| add XQDA metric learning on deep attributes | SSDAL · VIPeR | rank-1 accuracy: 37.9 -> 43.5 | Improved |
| add XQDA metric learning on deep attributes | SSDAL · GRID | rank-1 accuracy: 19.1 -> 22.4 | Improved |
| add XQDA metric learning on deep attributes | SSDAL · PRID | rank-1 accuracy: 20.1 -> 22.6 | Improved |
| fuse LOMO appearance features with attributes | SSDAL + XQDA · VIPeR | rank-1 accuracy: 43.5 -> 45.3 | Improved |
| use fine-tuned FC-7 features (T+U) instead of deep attributes (ablation) | SSDAL · VIPeR | rank-1 accuracy: 37.9 -> 27.4 | Hurt |
| fine-tune FC-7 on U only instead of T (ablation) | FC-7 features · VIPeR | rank-1 accuracy: 26.5 -> 10.1 | Hurt |
Closest papers by meaning
- Pose-Driven Deep Convolutional Model for Person Re-Identification (ICCV 2017 · similarity 0.90)
- Attribute-Guided Collaborative Learning for Partial Person Re-Identification (TPAMI 2023 · similarity 0.89)
- A Multi-Task Deep Network for Person Re-Identification (AAAI 2017 · similarity 0.89)
- Person30K: A Dual-Meta Generalization Network for Person Re-Identification (CVPR 2021 · similarity 0.89)
- Person Re-Identification by Camera Correlation Aware Feature Augmentation (TPAMI 2017 · similarity 0.89)
- Exploring the Camera Bias of Person Re-identification (ICLR 2025 · similarity 0.88)
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.
