Research map / Label-efficient and robust learning / Recognition and long-tail
Consistent-Aware Deep Learning for Person Re-Identification in a Camera Network
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 |
|---|---|---|---|
| Train the pretrained model with CADL instead of contrastive loss (ablation) | CADL person re-identification · Market-1501, camera-pair protocol (SQ) | weighted accuracy: 46.23 -> 60.09 | Helped |
| Train the pretrained model with CADL instead of contrastive loss (ablation) | CADL person re-identification · Market-1501, camera-pair protocol (MQ) | weighted accuracy: 59.16 -> 81.15 | Helped |
| Train with CADL instead of contrastive loss, then use cosine matching (ablation) | CADL person re-identification · Market-1501, standard protocol | rank-1 accuracy: 58.28 -> 73.84 | Helped |
| Train with CADL instead of contrastive loss, then use cosine matching (ablation) | CADL person re-identification · Market-1501, standard protocol | mAP: 41.28 -> 47.11 | Helped |
| Train with CADL instead of contrastive loss, then use cosine matching (ablation) | CADL person re-identification · RAiD, camera pairs (average) | rank-1 accuracy: 58.21 -> 77.38 | Helped |
| Train with CADL instead of contrastive loss, then use cosine matching (ablation) | CADL person re-identification · WARD, camera pairs (Cam12) | rank-1 accuracy: 63.57 -> 93.57 | Helped |
| Apply global association at test time after CADL training, rather than pairwise cosine matching (ablation) | CADL person re-identification · Market-1501, camera-pair protocol (SQ) | weighted accuracy: 52.88 -> 60.09 | Helped |
| Apply global association at test time after CADL training, rather than pairwise cosine matching (ablation) | CADL person re-identification · Market-1501, camera-pair protocol (MQ) | weighted accuracy: 72.02 -> 81.15 | Helped |
| Apply global association at test time after CADL training, rather than pairwise cosine matching (ablation) | CADL person re-identification · RAiD, camera pairs (average) | rank-1 accuracy: 77.38 -> 97.21 | Helped |
| Apply global association at test time after CADL training, rather than pairwise cosine matching (ablation) | CADL person re-identification · WARD, camera pairs (Cam12) | rank-1 accuracy: 93.57 -> 99.57 | Helped |
| Increase the inter-camera consistency weight µ from zero to a suitable value (ablation) | CADL association optimization · RAiD | average accuracy | Mixed |
| Increase the inter-camera consistency weight µ (ablation) | CADL association optimization · RAiD | variance across camera pairs | Helped |
Closest papers by meaning
- Deep Neural Networks with Inexact Matching for Person Re-Identification (NeurIPS 2016 · similarity 0.90)
- Group Consistent Similarity Learning via Deep CRF for Person Re-Identification (CVPR 2018 · similarity 0.90)
- Re-Identification With Consistent Attentive Siamese Networks (CVPR 2019 · similarity 0.89)
- Resource Aware Person Re-Identification Across Multiple Resolutions (CVPR 2018 · similarity 0.89)
- Efficient and Deep Person Re-Identification Using Multi-Level Similarity (CVPR 2018 · similarity 0.89)
- Harmonious Attention Network for Person Re-Identification (CVPR 2018 · 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.