Research map / Label-efficient and robust learning / Domain adaptation and generalization
Privacy-Preserving Adaptive Re-Identification without Image Transfer
Hamza Rami, Jhony H. Giraldo, Nicolas Winckler, Stéphane Lathuilière
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 MMD loss with Gaussian kernel to local client training (ablation) | Fed-Protoid · MS->Market | mAP: 42.6 -> 51.0 | Improved |
| use linear kernel MMD instead of no MMD (ablation) | Fed-Protoid · MS->Market | mAP: 42.6 -> 38.1 | Hurt |
| use order-2 kernel MMD instead of no MMD (ablation) | Fed-Protoid · MS->CUHK03-np | mAP: 22.4 -> 13.1 | Hurt |
| add teacher-student framework to the pseudo-client (ablation) | Fed-Protoid · MS->Market | mAP: 37.4 -> 51.0 | Improved |
| transmit teacher instead of student models for aggregation (ablation) | Fed-Protoid · MS->Market | mAP: 49.3 -> 51.0 | Improved |
| compute source prototypes with the global model instead of the pseudo-client model (ablation) | Fed-Protoid · MS->Market | mAP: 43.1 -> 51.0 | Improved |
| compute source prototypes with the global model (ViT variant) (ablation) | Fed-Protoid++ · MS->Market | mAP: 60.5 -> 61.7 | Improved |
| replace ResNet-50 with ViT-S (both ImageNet pretrained) (ablation) | Fed-Protoid · MS->CUHK03-np | mAP: 23.8 -> 27.5 | Improved |
| use ViT with LUPerson self-supervised pretraining instead of ResNet-50 ImageNet (ablation) | Fed-Protoid · MS->Market | mAP: 51 -> 61.7 | Improved |
| subsample source prototypes uniformly (ablation) | Fed-Protoid · RP->Market | mAP | No clear effect |
Closest papers by meaning
- Decentralised Learning from Independent Multi-Domain Labels for Person Re-Identification (AAAI 2021 · similarity 0.89)
- A Novel Unsupervised Camera-Aware Domain Adaptation Framework for Person Re-Identification (ICCV 2019 · similarity 0.89)
- Toward Real-World Holistic Privacy-Preserving Person Re-Identification (TPAMI 2026 · similarity 0.88)
- Alice Benchmarks: Connecting Real World Re-Identification with the Synthetic (ICLR 2024 · similarity 0.86)
- From Laboratory to Real World: A New Benchmark Towards Privacy-Preserved Visible-Infrared Person Re-Identification (CVPR 2025 · similarity 0.86)
- Person De-reidentification: A Variation-guided Identity Shift Modeling (CVPR 2025 · similarity 0.85)
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.
