Research map / Label-efficient and robust learning / Recognition and long-tail
Person Re-Identification With Cascaded Pairwise Convolutions
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 Channel Scaling layers (ablation) | BraidNet · CUHK03-Detected | Rank-1: 80.24 -> 83.32 | Helped |
| add Channel Scaling layers (ablation) | BraidNet · CUHK03-Labeled | Rank-1: 84.81 -> 87.29 | Helped |
| add Channel Scaling layers (ablation) | BraidNet · CUHK01 | Rank-1: 90.74 -> 89.14 | Hurt |
| replace Channel Scaling layers with Batch Normalization (ablation) | BraidNet-CS · CUHK01 | Rank-1: 89.14 -> 81.98 | Hurt |
| replace Channel Scaling layers with Batch Normalization (ablation) | BraidNet-CS · CUHK03-Detected | Rank-1: 83.32 -> 82.69 | Hurt |
| use SRL instead of fixed sample rate 1/4 (ablation) | BraidNet-CS · CUHK03-Detected | Rank-1: 83.32 -> 85.85 | Helped |
| use SRL instead of fixed sample rate 1/4 (ablation) | BraidNet-CS · CUHK03-Labeled | Rank-1: 87.29 -> 88.18 | Helped |
| use SRL instead of fixed sample rate 1/4 (ablation) | BraidNet-CS · CUHK01 | Rank-1: 89.14 -> 93.04 | Helped |
| vary the fixed sample rate among 1/8, 1/4, and 1/2 (ablation) | BraidNet-CS · CUHK03-Detected, CUHK03-Labeled, and CUHK01 | Rank-1: 82.82, 83.32, 83.63; 86.48, 87.29, 86.72; 88.87, 89.14, 90.78 | No clear effect |
| replace BraidNet's shared-image comparison with cascaded WConv asymmetric pattern matching | person re-identification models · CUHK03-Detected | Rank-1: +8.20 | Mixed |
Closest papers by meaning
- Recurrent Convolutional Network for Video-Based Person Re-Identification (CVPR 2016 · similarity 0.88)
- Person Re-Identification by Multi-Channel Parts-Based CNN With Improved Triplet Loss Function (CVPR 2016 · similarity 0.88)
- A Multi-Task Deep Network for Person Re-Identification (AAAI 2017 · similarity 0.88)
- Backbone Cannot Be Trained at Once: Rolling Back to Pre-Trained Network for Person Re-Identification (AAAI 2019 · similarity 0.88)
- Resource Aware Person Re-Identification Across Multiple Resolutions (CVPR 2018 · similarity 0.87)
- Joint Learning of Single-Image and Cross-Image Representations for Person Re-Identification (CVPR 2016 · 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.