Research map / Label-efficient and robust learning / Recognition and long-tail
Interpretable and Generalizable Person Re-Identification with Query-Adaptive Convolution and Temporal Lifting
Shengcai Liao, Ling Shao
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 |
|---|---|---|---|
| use QAConv matching instead of softmax cross-entropy (ablation) | ResNet-152 re-ID · Market-1501 -> DukeMTMC-reID | Rank-1 (%): 34.9 -> 54.4 | Improved |
| use QAConv matching instead of softmax cross-entropy (ablation) | ResNet-152 re-ID · DukeMTMC-reID -> Market-1501 | Rank-1 (%): 48.5 -> 62.8 | Improved |
| use QAConv matching instead of class memory loss with plain features (ablation) | ResNet-152 re-ID · Market-1501 -> DukeMTMC-reID | mAP (%): 21.8 -> 33.6 | Improved |
| add Arc loss margin instead of softmax (ablation) | ResNet-152 re-ID · Market-1501 -> DukeMTMC-reID | mAP (%): 18.4 -> 17.1 | No clear effect |
| add re-ranking (ablation) | QAConv · Market-1501 -> DukeMTMC-reID | mAP (%): 33.6 -> 52.4 | Improved |
| add re-ranking (ablation) | QAConv · DukeMTMC-reID -> Market-1501 | mAP (%): 31.6 -> 51.2 | Improved |
| add TLift on top of re-ranking (ablation) | QAConv + RR · Market-1501 -> DukeMTMC-reID | mAP (%): 52.4 -> 61.2 | Improved |
| add TLift on top of re-ranking (ablation) | QAConv + RR · DukeMTMC-reID -> Market-1501 | Rank-1 (%): 68.5 -> 78.7 | Improved |
| add TLift without re-ranking (ablation) | QAConv · DukeMTMC-reID -> Market-1501 | Rank-1 (%): 62.8 -> 61.5 | Mixed |
| use TFusion instead of TLift after re-ranking (ablation) | QAConv + RR · DukeMTMC-reID -> Market-1501 | mAP (%): 51.2 -> 47.2 | Hurt |
| use ResNet-152 backbone instead of ResNet-50 (ablation) | QAConv · Market-1501 -> DukeMTMC-reID | Rank-1 (%): 48.8 -> 54.4 | Improved |
| use ResNet-152 backbone instead of ResNet-50 (ablation) | QAConv · DukeMTMC-reID -> Market-1501 | mAP (%): 27.2 -> 31.6 | Improved |
Closest papers by meaning
- Deep Neural Networks with Inexact Matching for Person Re-Identification (NeurIPS 2016 · similarity 0.93)
- Person Re-Identification by Camera Correlation Aware Feature Augmentation (TPAMI 2017 · similarity 0.90)
- Pose-Driven Deep Convolutional Model for Person Re-Identification (ICCV 2017 · similarity 0.90)
- Multi-Scale Deep Learning Architectures for Person Re-Identification (ICCV 2017 · similarity 0.90)
- Joint Learning of Single-Image and Cross-Image Representations for Person Re-Identification (CVPR 2016 · similarity 0.90)
- Cheb-GR: Rethinking K-nearest Neighbor Search in Re-ranking for Person Re-identification (CVPR 2025 · 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.
