Research map / Label-efficient and robust learning / Recognition and long-tail
A Systematic Evaluation and Benchmark for Person Re-Identification: Features, Metrics, and Datasets
Srikrishna Karanam, Mengran Gou, Ziyan Wu, Angels Rates-Borras, Octavia Camps, Richard J. Radke
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (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 |
|---|---|---|---|
| use multi-shot ranking SRID instead of feature averaging AVER (ablation) | GOG-KISSME · multi-shot datasets | rank-1 | Improved |
| add metric learning to multi-shot ranking (ablation) | multi-shot ranking · multi-shot datasets | rank-1 | Improved |
| apply PCA to 100 dimensions on GOG (ablation) | GOG · single-shot datasets | rank-1 | Hurt |
| increase number of horizontal strips from 6 to 9 or 15 (ablation) | GOG · single- and multi-shot datasets | rank-1 | No clear effect |
| increase PCA dimension from 50 to 500 (ablation) | GOG · VIPeR and Market1501 | rank-1 and training time | Mixed |
| use IDE-ResNet instead of other features with NFST_exp | NFST_exp · 7 single-shot datasets | rank-1: best on 4 of 7 datasets | Mixed |
| use GOG instead of other features with SRID and KISSME | SRID-KISSME · 11 multi-shot datasets | rank-1: best on all 11 datasets | Improved |
| use KISSME instead of other metric learners with GOG and SRID | GOG-SRID · 11 multi-shot datasets | rank-1: best across all datasets | Improved |
| use IDE-ResNet instead of IDE-CaffeNet | IDE · single-shot, variation VV | mean rank-1: 29.2 -> 40.7 | Improved |
| use IDE-ResNet instead of GOG | feature · single-shot, variation RES | mean rank-1: 19.5 -> 38.3 | Improved |
| use GOG instead of IDE-ResNet | feature · multi-shot, variation RES | mean rank-1: 34.0 -> 28.4 | Hurt |
| use IDE-ResNet instead of IDE-VGG | IDE · multi-shot, variation IV | mean rank-1: 61.5 -> 65.1 | Improved |
Closest papers by meaning
- Deep Learning for Person Re-Identification: A Survey and Outlook (TPAMI 2021 · similarity 0.94)
- Person Re-Identification in the Wild (CVPR 2017 · similarity 0.93)
- Revisiting Person Re-Identification by Camera Selection (TPAMI 2023 · similarity 0.91)
- Person Re-Identification by Camera Correlation Aware Feature Augmentation (TPAMI 2017 · similarity 0.91)
- Video-Based Person Re-Identification via Self Paced Weighting (AAAI 2018 · similarity 0.90)
- A Bottom-Up Clustering Approach to Unsupervised Person Re-Identification (AAAI 2019 · similarity 0.90)
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.
