Research map / Detection and segmentation / Open-set and few-shot detection
Cross-Domain Detection via Graph-Induced Prototype Alignment
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 |
|---|---|---|---|
| Apply GPA with two-stage rather than single-stage alignment (ablation) | Faster R-CNN cross-domain detection · Foggy Cityscapes val (Cityscapes → Foggy Cityscapes) | mAP: 37.4 -> 39.5; 37.6 -> 39.5 | Helped |
| Apply GPA with two-stage rather than single-stage alignment (ablation) | Faster R-CNN cross-domain detection · Cityscapes val (SIM 10k → Cityscapes) | car AP: 45.1 -> 47.6; 44.8 -> 47.6 | Helped |
| Apply GPA with two-stage rather than single-stage alignment (ablation) | Faster R-CNN cross-domain detection · Cityscapes val (KITTI → Cityscapes) | car AP: 46.9 -> 47.9; 46.1 -> 47.9 | Helped |
| Use IoU rather than Euclidean distance to construct the relation graph (ablation) | GPA · Cityscapes val (SIM 10k → Cityscapes) | car AP: 46.1 -> 47.6 | Helped |
| Add a learnable parameter matrix to Euclidean-distance graph propagation (ablation) | GPA · Cityscapes val (SIM 10k → Cityscapes) | car AP: 46.1 -> 43.2 | Hurt |
| Add a learnable parameter matrix to IoU graph propagation (ablation) | GPA · Cityscapes val (SIM 10k → Cityscapes) | car AP: 47.6 -> 43.6 | Hurt |
| Add relation-graph aggregation rather than using original RPN proposals directly (ablation) | GPA · Cityscapes val (SIM 10k → Cityscapes) | car AP: 45.0 -> 47.6 | Helped |
| Use GPA instead of the strongest listed baseline | Faster R-CNN cross-domain detection · Foggy Cityscapes val (Cityscapes → Foggy Cityscapes) | mAP: 35.9 -> 39.5 | Helped |
| Use GPA instead of the strongest listed baseline | Faster R-CNN cross-domain detection · Cityscapes val (SIM 10k → Cityscapes) | car AP: 46.6 -> 47.6 | Helped |
| Use GPA instead of the strongest listed baseline | Faster R-CNN cross-domain detection · Cityscapes val (KITTI → Cityscapes) | car AP: 43.6 -> 47.9 | Helped |
| Vary the two-stage alignment loss trade-off parameters within the reported range (ablation) | GPA · Cityscapes val (SIM 10k → Cityscapes) | car AP | No clear effect |
| Set the class-balancing parameter γ around 2.0 (ablation) | GPA · Foggy Cityscapes (Cityscapes → Foggy Cityscapes) | mAP | Helped |
Closest papers by meaning
- RPN Prototype Alignment for Domain Adaptive Object Detector (CVPR 2021 · similarity 0.91)
- Adapting Object Detectors via Selective Cross-Domain Alignment (CVPR 2019 · similarity 0.90)
- SCAN: Cross Domain Object Detection with Semantic Conditioned Adaptation (AAAI 2022 · similarity 0.90)
- Universal Domain Adaptive Object Detection via Dual Probabilistic Alignment (AAAI 2025 · similarity 0.89)
- Mean Teacher DETR with Masked Feature Alignment: A Robust Domain Adaptive Detection Transformer Framework (AAAI 2024 · similarity 0.89)
- Improving Single Domain-Generalized Object Detection: A Focus on Diversification and Alignment (CVPR 2024 · 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.