Research map / Label-efficient and robust learning / Recognition and long-tail
A Multi-Task Deep Network for Person Re-Identification
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| jointly optimize classification and ranking tasks in one network (ablation) | MTDnet · CUHK03 | rank-1 accuracy: 68.35 -> 74.68; 60.13 -> 74.68 | Helped |
| jointly optimize classification and ranking tasks in one network (ablation) | MTDnet · CUHK01, VIPeR, iLIDS, PRID2011 | rank-1 accuracy: 76.50 -> 77.50; 44.30 -> 45.89; 54.34 -> 57.80; 28.00 -> 32.00 | Helped |
| add cross-domain feature-map alignment using CUHK03 source data (ablation) | MTDnet · CUHK01 | rank-1 accuracy: 76.50 -> 78.50 | Helped |
| add cross-domain feature-map alignment using CUHK03 source data (ablation) | MTDnet · VIPeR | rank-1 accuracy: 44.30 -> 47.47 | Helped |
| add cross-domain feature-map alignment using CUHK03 source data (ablation) | MTDnet · iLIDS | rank-1 accuracy: 54.34 -> 58.38 | Helped |
| add cross-domain feature-map alignment using CUHK03 source data (ablation) | MTDnet · PRID2011 | rank-1 accuracy: 32.00 -> 31.00 | Hurt |
| combine source and target datasets directly for training (ablation) | MTDnet · CUHK01, VIPeR, iLIDS, PRID2011 | rank-1 accuracy: 77.50 -> 75.50; 45.89 -> 43.35; 57.80 -> 54.91; 32.00 -> 27.00 | Hurt |
| test the multi-task model with ranking loss rather than its classification architecture (ablation) | MTDnet · CUHK03, CUHK01, VIPeR, iLIDS, PRID2011 | rank-1 accuracy: 74.68 -> 66.03; 77.50 -> 66.00; 45.89 -> 34.81; 57.80 -> 46.82; 32.00 -> 26.00 | Hurt |
Closest papers by meaning
- Backbone Cannot Be Trained at Once: Rolling Back to Pre-Trained Network for Person Re-Identification (AAAI 2019 · similarity 0.92)
- Single Camera Training for Person Re-Identification (AAAI 2020 · similarity 0.91)
- Interaction-And-Aggregation Network for Person Re-Identification (CVPR 2019 · similarity 0.90)
- Resource Aware Person Re-Identification Across Multiple Resolutions (CVPR 2018 · similarity 0.90)
- Efficient and Deep Person Re-Identification Using Multi-Level Similarity (CVPR 2018 · similarity 0.90)
- Instruct-ReID: A Multi-purpose Person Re-identification Task with Instructions (CVPR 2024 · 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.