Research map / Label-efficient and robust learning / Recognition and long-tail
A Dual-Path Model With Adaptive Attention for Vehicle Re-Identification
Pirazh Khorramshahi, Amit Kumar, Neehar Peri, Sai Saketh Rambhatla, Jun-Cheng Chen, Rama Chellappa
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 the orientation-conditioned key-point attention branch to ResNet-101 | ResNet-101 baseline · VeRi-776 | mAP: 55.75 -> 61.18 | Improved |
| add the orientation-conditioned key-point attention branch to ResNet-50 | ResNet-50 baseline · VeRi-776 | mAP: 52.88 -> 58.52 | Improved |
| add the attention branch to ResNet-101 | ResNet-101 baseline · VehicleID (test size 800) | CMC@1: 70.03 -> 74.69 | Improved |
| add the attention branch to ResNet-101 | ResNet-101 baseline · VehicleID (test size 2400) | CMC@1: 59.04 -> 63.54 | Improved |
| add the attention branch to ResNet-101 | ResNet-101 baseline · VeRi-Wild (Large) | mAP: 38.99 -> 41.68 | Improved |
| add re-ranking post-processing | AAVER (ResNet-101) · VeRi-776 | mAP: 61.18 -> 66.35 | Improved |
| add two-branch design instead of single branch with heatmaps concatenated (ablation) | AAVER (ResNet-101) · VeRi-776 | mAP: 0.528 -> 0.612 | Improved |
| add two-branch design instead of single branch (ablation) | AAVER (ResNet-101) · VehicleID (test size 800) | CMC@1: 69.61 -> 74.69 | Improved |
| add adaptive key-point selection instead of combining all heatmaps (ablation) | AAVER (ResNet-101) · VeRi-776 | CMC@1: 87.66 -> 88.97 | Improved |
| add adaptive key-point selection instead of combining all heatmaps (ablation) | AAVER (ResNet-101) · VehicleID (test size 800) | CMC@1: 71.79 -> 74.69 | Improved |
| pool shallow (56x56, Res2) rather than deeper (14x14) global features (ablation) | AAVER (ResNet-101) · VeRi-776 | CMC@1: 85.88 -> 88.97 | Improved |
| pool shallow (56x56) rather than 28x28 global features (ablation) | AAVER (ResNet-101) · VeRi-776 | mAP: 0.608 -> 0.612 | Improved |
Closest papers by meaning
- The Devil is in the Details: Self-Supervised Attention for Vehicle Re-Identification (ECCV 2020 · similarity 0.89)
- Infer the Whole from a Glimpse of a Part: Keypoint-Based Knowledge Graph for Vehicle Re-Identification (AAAI 2025 · similarity 0.87)
- Viewpoint-Aware Attentive Multi-View Inference for Vehicle Re-Identification (CVPR 2018 · similarity 0.86)
- Learning Deep Neural Networks for Vehicle Re-ID With Visual-Spatio-Temporal Path Proposals (ICCV 2017 · similarity 0.86)
- CLIP-driven View-aware Prompt Learning for Unsupervised Vehicle Re-identification (AAAI 2025 · similarity 0.85)
- Alice Benchmarks: Connecting Real World Re-Identification with the Synthetic (ICLR 2024 · similarity 0.83)
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.
