Research map / Label-efficient and robust learning / Recognition and long-tail
Deep Spatial Feature Reconstruction for Partial Person Re-Identification: Alignment-Free Approach
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 multi-scale blocks (1×1, 2×2, and 3×3) to DSR (ablation) | DSR with FCN · Partial REID single-shot | rank-1 accuracy: 39.33 -> 43.00 | Helped |
| add multi-scale blocks (1×1, 2×2, and 3×3) to DSR (ablation) | DSR with FCN · Partial-iLIDS single-shot | rank-1 accuracy: 51.06 -> 54.58 | Helped |
| add multi-scale blocks (1×1, 2×2, and 3×3) to DSR (ablation) | DSR with FCN · Partial REID multi-shot (N=3) | rank-1 accuracy: 49.33 -> 53.67 | Helped |
| add multi-scale blocks (1×1, 2×2, and 3×3) to DSR (ablation) | DSR with FCN · Partial-iLIDS multi-shot (N=3) | rank-1 accuracy: 54.67 -> 55.46 | Helped |
| add multi-scale blocks to DSR (ablation) | ResNet50-res5c+DSR · Market1501 single query | rank-1 accuracy: 82.72 -> 83.58 | Helped |
| add multi-scale blocks to DSR (ablation) | ResNet50-res5c+DSR · Market1501 single query | mAP: 61.25 -> 64.25 | Helped |
| fine-tune the pre-trained FCN using DSR verification loss (ablation) | FCN · Partial REID single-shot | CMC performance | Helped |
| use three gallery images per identity instead of one | DSR (single-scale) · Partial REID | rank-1 accuracy: 39.33 -> 49.33 | Helped |
| use three gallery images per identity instead of one | DSR (single-scale) · Partial-iLIDS | rank-1 accuracy: 51.06 -> 54.67 | Helped |
| avoid resizing partial images to a fixed size and use DSR matching | Partial person re-identification · Partial REID single-shot | rank-1 accuracy: 19.33 -> 39.33 | Helped |
| avoid resizing partial images to a fixed size and use DSR matching | Partial person re-identification · Partial-iLIDS single-shot | rank-1 accuracy: 21.85 -> 51.06 | Helped |
Closest papers by meaning
- Deep Neural Networks with Inexact Matching for Person Re-Identification (NeurIPS 2016 · similarity 0.90)
- Similarity Learning With Spatial Constraints for Person Re-Identification (CVPR 2016 · similarity 0.89)
- Efficient and Deep Person Re-Identification Using Multi-Level Similarity (CVPR 2018 · similarity 0.89)
- Spatial-Temporal Graph Convolutional Network for Video-Based Person Re-Identification (CVPR 2020 · similarity 0.88)
- Matching on Sets: Conquer Occluded Person Re-identification Without Alignment (AAAI 2021 · similarity 0.88)
- Keypoint Message Passing for Video-Based Person Re-identification (AAAI 2022 · similarity 0.88)
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.