Research map / Architectures and efficiency / Neural architecture search
Combined Depth Space Based Architecture Search for Person Re-Identification
Check 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 |
|---|---|---|---|
| Add adaptive depth through CDBlocks | CDNet versus CNet · Market1501 | mAP: 83.5 -> 83.7 | Helped |
| Use top-2 rather than top-1 branch sampling during search | CDNet · Market1501 search evaluation | mAP: 83.2 -> 83.7 | Helped |
| Increase the number of sampled branches beyond top-2 | CDNet search · Market1501 search evaluation | mAP: 83.7 -> 83.5 -> 83.0 | Hurt |
| Add FNeck local features using 2 partitions (ablation) | CDNet · Market1501 | mAP: 81.5 -> 83.7 | Helped |
| Increase FNeck partitions from 2 to 3 (ablation) | CDNet · Market1501 | mAP: 83.7 -> 82.7 | Hurt |
| Use 4 rather than 2 FNeck partitions (ablation) | CDNet · Market1501 | rank-1: 93.7 -> 94.2 | Helped |
| Replace BNNeck with BLNeck (ablation) | CDNet body · Market1501 | mAP: 77.6 -> 81.1 | Helped |
| Add FNeck to BLNeck to form FBLNeck (ablation) | CDNet body · Market1501 | mAP: 81.1 -> 83.7 | Helped |
| Replace the two-branch CBlock with a single branch (ablation) | CDNet · Market1501 | mAP: 83.7 -> 79.7 | Hurt |
| Fix CBlock kernel combinations to (3,5) (ablation) | CDNet · Market1501 | mAP: 83.7 -> 82.0 | Hurt |
| Attach FBLNeck (ablation) | OSNet · Market1501 | mAP: 82.2 -> 83.4 | Helped |
| Reduce image resolution while keeping width multiplier at 1.0 (ablation) | CDNet · Market1501 | mAP: 86.0 -> 82.3 | Hurt |
Closest papers by meaning
- A Multi-Task Deep Network for Person Re-Identification (AAAI 2017 · similarity 0.88)
- Relation Network for Person Re-Identification (AAAI 2020 · similarity 0.88)
- Backbone Cannot Be Trained at Once: Rolling Back to Pre-Trained Network for Person Re-Identification (AAAI 2019 · similarity 0.87)
- Single Camera Training for Person Re-Identification (AAAI 2020 · similarity 0.87)
- Person30K: A Dual-Meta Generalization Network for Person Re-Identification (CVPR 2021 · similarity 0.87)
- Instruct-ReID: A Multi-purpose Person Re-identification Task with Instructions (CVPR 2024 · 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.