Research map / Label-efficient and robust learning / Domain adaptation and generalization
Meta Batch-Instance Normalization for Generalizable Person Re-Identification
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 |
|---|---|---|---|
| Generalize learnable BN-IN balancing parameters through meta-learning simulations of failed normalization (ablation) | MetaBIN · Large-scale DG Re-ID benchmark (average Rank-1) | Rank-1: 50.9 -> 64.7 | Helped |
| Use triplet loss rather than cross-entropy loss for meta-train and meta-test (ablation) | MetaBIN · Large-scale DG Re-ID benchmark (average Rank-1) | Rank-1: 60.6 -> 62.8 | Helped |
| Add intra-domain scatter loss to triplet meta-train loss (ablation) | MetaBIN · Large-scale DG Re-ID benchmark (average Rank-1) | Rank-1: 62.8 -> 63.0 | Helped |
| Add inter-domain shuffle loss to triplet meta-train loss (ablation) | MetaBIN · Large-scale DG Re-ID benchmark (average Rank-1) | Rank-1: 62.8 -> 63.1 | Helped |
| Use both scatter and shuffle losses with triplet meta-train loss (ablation) | MetaBIN · Large-scale DG Re-ID benchmark (average Rank-1) | Rank-1: 62.8 -> 63.5 | Helped |
| Use cyclic rather than fixed inner step size (ablation) | MetaBIN · Large-scale DG Re-ID benchmark (average Rank-1) | Rank-1: 63.5 -> 64.7 | Helped |
| Remove separation between base-model and balancing-parameter update episodes (ablation) | MetaBIN · Large-scale DG Re-ID benchmark (average Rank-1) | Rank-1: 64.7 -> 60.9 | Hurt |
| Replace MetaBIN normalization with BIN (ablation) | MetaBIN · Large-scale DG Re-ID benchmark (average Rank-1) | Rank-1: 64.7 -> 60.6 | Hurt |
| Increase the weight of shuffle and triplet meta-train losses (ablation) | MetaBIN · Large-scale DG Re-ID benchmark | performance | Hurt |
| Use MetaBIN instead of SNR | ResNet-50 cross-domain Re-ID · Market-1501 → DukeMTMC | mAP: 33.6 -> 33.1 | Hurt |
| Use MetaBIN instead of SNR | ResNet-50 cross-domain Re-ID · DukeMTMC → Market-1501 | mAP: 33.9 -> 35.9 | Helped |
Closest papers by meaning
- Generalizable Person Re-identification via Self-Supervised Batch Norm Test-Time Adaption (AAAI 2022 · similarity 0.93)
- Learning to Generalize Unseen Domains via Memory-based Multi-Source Meta-Learning for Person Re-Identification (CVPR 2021 · similarity 0.92)
- Person30K: A Dual-Meta Generalization Network for Person Re-Identification (CVPR 2021 · similarity 0.92)
- Style Normalization and Restitution for Generalizable Person Re-Identification (CVPR 2020 · similarity 0.91)
- Generalizable Person Re-identification via Balancing Alignment and Uniformity (NeurIPS 2024 · similarity 0.88)
- Diversity-Authenticity Co-constrained Stylization for Federated Domain Generalization in Person Re-identification (AAAI 2024 · similarity 0.87)
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.