Research map / Label-efficient and robust learning / Domain adaptation and generalization
Learning to Adapt Invariance in Memory for Person Re-identification
Zhun Zhong, Liang Zheng, Zhiming Luo, Shaozi Li, Yi Yang
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 |
|---|---|---|---|
| add exemplar-invariance learning (ablation) | Source only model · Duke->Market-1501 | Rank-1: 43.1 -> 48.7 | Improved |
| add exemplar-invariance learning (ablation) | Source only model · Market-1501->Duke | Rank-1: 28.9 -> 34.2 | Improved |
| add camera-invariance to EI model (ablation) | Ours w/ EI · Duke->Market-1501 | Rank-1: 48.7 -> 63.1 | Improved |
| add neighborhood-invariance to EI model (ablation) | Ours w/ EI · Market-1501->Duke | Rank-1: 34.2 -> 67.2 | Improved |
| add neighborhood-invariance to EI+CI (ablation) | Ours w/ EI+CI · Duke->Market-1501 | Rank-1: 63.1 -> 84.1 | Improved |
| add camera-invariance to EI+NI (ablation) | Ours w/ EI+NI · Duke->Market-1501 | Rank-1: 71.8 -> 84.1 | Improved |
| use exemplar memory instead of mini-batch invariance learning (ablation) | Ours (VNS) · Duke->Market-1501 | Rank-1: 70.2 -> 77.4 | Improved |
| add graph-based positive prediction over vanilla neighbor selection (ablation) | Ours with memory · Duke->Market-1501 | mAP: 45.5 -> 63.8 | Improved |
| threshold selection (p>=0.9) instead of top-8 with GPP (ablation) | GPP · Duke->Market-1501 | Rank-1: 79.8 -> 84.1 | Improved |
| threshold selection on classifier without graph convolution (ablation) | VNS · Duke->Market-1501 | Rank-1: 77.4 -> 77.0 | No clear effect |
| set temperature beta to 0.05 instead of 0.1 (ablation) | Ours · Duke->Market-1501 | Rank-1: 75.7 -> 84.1 | Improved |
| set temperature beta to 0.05 instead of 0.01 (ablation) | Ours · Duke->Market-1501 | Rank-1: 50.1 -> 84.1 | Improved |
Closest papers by meaning
- Invariance Matters: Exemplar Memory for Domain Adaptive Person Re-Identification (CVPR 2019 · similarity 0.96)
- Generalizing Person Re-Identification by Camera-Aware Invariance Learning and Cross-Domain Mixup (ECCV 2020 · similarity 0.93)
- Joint Disentangling and Adaptation for Cross-Domain Person Re-Identification (ECCV 2020 · similarity 0.91)
- A Novel Unsupervised Camera-Aware Domain Adaptation Framework for Person Re-Identification (ICCV 2019 · similarity 0.91)
- Generalizing A Person Retrieval Model Hetero- and Homogeneously (ECCV 2018 · similarity 0.91)
- Towards Discriminative Representation Learning for Unsupervised Person Re-Identification (ICCV 2021 · similarity 0.91)
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.
