Research map / Detection and segmentation / Open-set and few-shot detection
Relation Matters: Foreground-aware Graph-based Relational Reasoning for Domain Adaptive Object Detection
Chaoqi Chen, Jiongcheng Li, Hong-Yu Zhou, Xiaoguang Han, Yue Huang, Xinghao Ding, Yizhou Yu
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 |
|---|---|---|---|
| upgrade DBGL to FGRR (foreground-aware relational reasoning) | DBGL · Pascal VOC to Clipart | mAP: 41.6% -> 43.3% | Improved |
| upgrade DBGL to FGRR | DBGL · Pascal VOC to Watercolor2k | mAP: 53.8% -> 55.7% | Improved |
| upgrade DBGL to FGRR | DBGL · Pascal VOC to Comic2k | mAP: 29.7% -> 32.7% | Improved |
| upgrade DBGL to FGRR | DBGL · Public to In-house mammogram | AP on mass: 38.4 -> 43.9 | Improved |
| model intra and inter-domain relations instead of MTOR object relation consistency | MTOR · Cityscapes to Foggy-Cityscapes | mAP: 35.1% -> 40.8% | Improved |
| use relational reasoning instead of prototype alignment (C2F, GPA) | C2F and GPA · Cityscapes to Foggy-Cityscapes | mAP: 38.6% and 39.5% -> 40.8% | Improved |
| add FPN to the detection backbone | DAOD methods · Sim10K to Cityscapes | AP on car: 44.5 -> 55.8 | Improved |
| apply FGRR adaptation versus no adaptation | Faster-RCNN ResNet-101 · Public to In-house mammogram | AP on mass: 12.0 -> 43.9 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 0 of 1 on most shared results.
- MeGA-CDA — MeGA-CDA: Memory Guided Attention for Category-Aware Unsupervised Domain Adaptive Object Detection (CVPR 2021) · better on 0 of 2 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Universal Domain Adaptive Object Detection via Dual Probabilistic Alignment (AAAI 2025 · similarity 0.90)
- SEEN-DA: SEmantic ENtropy guided Domain-aware Attention for Domain Adaptive Object Detection (CVPR 2025 · similarity 0.90)
- DA-Ada: Learning Domain-Aware Adapter for Domain Adaptive Object Detection (NeurIPS 2024 · similarity 0.89)
- Differential Alignment for Domain Adaptive Object Detection (AAAI 2025 · similarity 0.88)
- Triple Feature Disentanglement for One-Stage Adaptive Object Detection (AAAI 2024 · similarity 0.87)
- DSD-DA: Distillation-based Source Debiasing for Domain Adaptive Object Detection (ICML 2024 · 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.
