Research map / Label-efficient and robust learning / Recognition and long-tail
Unsupervised Person Re-Identification by Camera-Aware Similarity Consistency Learning
Ancong Wu, Wei-Shi Zheng, Jian-Huang Lai
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 Lcon alone to pretrained model (ablation) | pretrained ResNet-50 · Market-1501 / DukeMTMC | Rank-1 (%): 51.5 -> 54.4 (Market); 47.6 -> 40.6 (Duke) | Mixed |
| add Lpre alone to pretrained model (ablation) | pretrained ResNet-50 · Market-1501 | Rank-1 (%): 51.5 -> 51.2 | No clear effect |
| add Lpre to Lcon (joint loss) (ablation) | Lcon · DukeMTMC | Rank-1 (%): 40.6 -> 56.6 | Improved |
| add Lpre to Lcon (joint loss) (ablation) | Lcon · Market-1501 | Rank-1 (%): 54.4 -> 61.4 | Improved |
| use Log-Euclidean metric instead of Euclidean in Lpre (ablation) | Lpre + Lcon · Market-1501 | Rank-1 (%): 59.2 -> 61.4 | Improved |
| use Log-Euclidean metric instead of Euclidean in Lpre (ablation) | Lpre + Lcon · DukeMTMC | mAP (%): 36.2 -> 35.8 | Hurt |
| add fine consistency learning (step 2) (ablation) | Lpre + Lcon (step 1) · Market-1501 | Rank-1 (%): 61.4 -> 65.4 | Improved |
| add fine consistency learning (step 2) (ablation) | Lpre + Lcon (step 1) · DukeMTMC | Rank-1 (%): 56.6 -> 59.3 | Improved |
| apply MMD feature alignment to pretrained model | pretrained ResNet-50 · Market-1501 | Rank-1 (%): 51.5 -> 28.1 | Hurt |
| apply CORAL feature alignment to pretrained model | pretrained ResNet-50 · DukeMTMC | Rank-1 (%): 47.6 -> 14.0 | Hurt |
| add Lpre to MMD alignment (ablation) | MMD · Market-1501 | Rank-1 (%): 28.1 -> 58.7 | Improved |
| use Lcon instead of MMD alongside Lpre (ablation) | Lpre · Market-1501 | Rank-1 (%): 58.7 -> 61.4 | Improved |
Closest papers by meaning
- Intra-Inter Camera Similarity for Unsupervised Person Re-Identification (CVPR 2021 · similarity 0.92)
- Intra-Inter Domain Similarity for Unsupervised Person Re-Identification (TPAMI 2022 · similarity 0.92)
- Cross-View Asymmetric Metric Learning for Unsupervised Person Re-Identification (ICCV 2017 · similarity 0.91)
- Camera-Aware Proxies for Unsupervised Person Re-Identification (AAAI 2021 · similarity 0.90)
- Unsupervised Domain Adaptation for Person Re-identification via Heterogeneous Graph Alignment (AAAI 2021 · similarity 0.90)
- CDE-Learning: Camera Deviation Elimination Learning for Unsupervised Person Re-identification (AAAI 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.
