Research map / Label-efficient and robust learning / Recognition and long-tail
Large Scale Similarity Learning Using Similar Pairs for Person Verification
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 bilinear similarity metric Mb to Mahalanobis metric Md (joint metrics) (ablation) | LSSL · LFW restricted | accuracy (%): 82.6 -> 85.5 | Helped |
| add bilinear similarity metric Mb to Mahalanobis metric Md (joint metrics) (ablation) | LSSL · PubFig restricted | accuracy (%): 78.3 -> 79.2 | Helped |
| use Mahalanobis metric instead of bilinear similarity metric alone (ablation) | LSSL · PubFig restricted | accuracy (%): 73.9 -> 78.3 | Helped |
| derive dissimilar-pair prior from similar pairs (iKISSME) instead of estimating from dissimilar pairs | KISSME · LFW restricted | accuracy (%): 80.6 -> 82.2 | Helped |
| derive dissimilar-pair prior from similar pairs (iKISSME) | KISSME · PubFig restricted | accuracy (%): 77.7 -> 78.6 | Helped |
| use Intra-PCA instead of PCA for feature reduction (ablation) | LSSL · LFW restricted | accuracy (%): 85.5 -> 86.2 | Helped |
| use Intra-PCA for KISSME | KISSME · LFW restricted | accuracy (%) | Hurt |
| learn from similar pairs only, skipping dissimilar pairs | LSSL vs KISSME · VIPeR | training time (ms): 66.8 ms -> 1.6 ms | Helped |
| LSSL vs best prior method MLF | LSSL · VIPeR | rank-1 matching rate (%): 43.4% -> 47.8% | Helped |
Closest papers by meaning
- Similarity Learning With Spatial Constraints for Person Re-Identification (CVPR 2016 · similarity 0.88)
- Group Consistent Similarity Learning via Deep CRF for Person Re-Identification (CVPR 2018 · similarity 0.86)
- Semi-Supervised Bayesian Attribute Learning for Person Re-Identification (AAAI 2018 · similarity 0.86)
- PairDETR : Joint Detection and Association of Human Bodies and Faces (CVPR 2024 · similarity 0.84)
- CA-Jaccard: Camera-aware Jaccard Distance for Person Re-identification (CVPR 2024 · similarity 0.84)
- Vec2Face: Scaling Face Dataset Generation with Loosely Constrained Vectors (ICLR 2025 · similarity 0.83)
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.