Research map / Label-efficient and robust learning / Recognition and long-tail
Learning Incremental Triplet Margin 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 |
|---|---|---|---|
| Add incremental triplet-margin feature shifting | TriNet-S · Market-1501 | mAP: 77.9 -> 82.3 | Helped |
| Add global hard identity searching to LITM | LITM · Market-1501 | mAP: 82.3 -> 83.9 | Helped |
| Add global hard identity searching to LITM | LITM · DukeMTMC-reID | Rank-1: 84.8 -> 85.9 | Helped |
| Replace GAP with GMP (ablation) | TriNet · Market-1501 | mAP: 71.6 -> 75.9 | Helped |
| Make the network fully convolutional by removing trailing fully connected layers (ablation) | TriNet · Market-1501 | mAP: 70.6 -> 71.6 | Helped |
| Reduce conv5_x stride from 2 to 1 to increase feature-map resolution (ablation) | TriNet-S · Market-1501 | mAP: 75.9 -> 77.9 | Helped |
| Increase one-stage triplet margin from 4 to 7 (ablation) | TriNet-S · Market-1501 | mAP: 79.1 -> 78.2 | Hurt |
| Increase one-stage triplet margin from 7 to 10 (ablation) | TriNet-S · Market-1501 | mAP: 78.2 -> 77.7 | Hurt |
| Add one feature-shift iteration with increased triplet margin (ablation) | LITM · Market-1501 | mAP: 80.9 -> 82.2 | Helped |
| Add a second feature-shift iteration (ablation) | LITM · Market-1501 | mAP: 82.2 -> 82.3 | Helped |
| Use mid-level feature maps for feature shifts instead of conv5_x at every stage (ablation) | LITM · Market-1501 | mAP: 81.2 -> 82.3 | Helped |
| Use low-level conv3_x features for the base embedding and high-level features for later shifts (ablation) | LITM · Market-1501 | mAP: 82.3 -> 79.4 | Hurt |
Closest papers by meaning
- Beyond Triplet Loss: A Deep Quadruplet Network for Person Re-Identification (CVPR 2017 · similarity 0.91)
- From Poses to Identity: Training-Free Person Re-Identification via Feature Centralization (CVPR 2025 · similarity 0.87)
- Instruct-ReID: A Multi-purpose Person Re-identification Task with Instructions (CVPR 2024 · similarity 0.87)
- CLIMB-ReID: A Hybrid CLIP-Mamba Framework for Person Re-Identification (AAAI 2025 · similarity 0.87)
- Exploring the Camera Bias of Person Re-identification (ICLR 2025 · similarity 0.86)
- Cheb-GR: Rethinking K-nearest Neighbor Search in Re-ranking for Person Re-identification (CVPR 2025 · 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.