Research map / Detection and segmentation / Open-set and few-shot detection
Instance Relation Graph Guided Source-Free Domain Adaptive Object Detection
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 IRG-guided contrastive learning to the mean-teacher baseline (ablation) | Source-free detector adaptation · Cityscapes → FoggyCityscapes | mAP: 34.3 -> 37.1 | Helped |
| Use strong student augmentation and weak teacher augmentation instead of weak-weak augmentation (ablation) | Mean-teacher adaptation · Cityscapes → FoggyCityscapes | mAP: 32.2 -> 34.3 | Helped |
| Use strong student augmentation and weak teacher augmentation instead of strong-strong augmentation (ablation) | Mean-teacher adaptation · Cityscapes → FoggyCityscapes | mAP: 29.9 -> 34.3 | Helped |
| Add graph distillation loss (ablation) | Mean-teacher adaptation with strong-weak augmentations · Cityscapes → FoggyCityscapes | mAP: 34.3 -> 35.9 | Helped |
| Add graph-guided contrastive learning (ablation) | IRG with graph distillation loss · Cityscapes → FoggyCityscapes | mAP: 35.9 -> 37.1 | Helped |
| Add supervised contrastive learning across student and teacher proposals (ablation) | Mean-teacher adaptation · Cityscapes → FoggyCityscapes | mAP: 34.3 -> 36.2 | Helped |
| Add graph distillation loss to the supervised-contrastive mean-teacher setup (ablation) | Mean-teacher adaptation with SimCLR contrastive loss · Cityscapes → FoggyCityscapes | mAP: 36.2 -> 35.9 | Hurt |
| Replace learnable IRG proposal relations with K-means clustering (ablation) | Graph contrastive proposal learning · Cityscapes → FoggyCityscapes | mAP: 37.1 -> 36.5 | Hurt |
| Freeze IRG and use original edge weights (ablation) | Graph contrastive proposal learning · Cityscapes → FoggyCityscapes | mAP: 37.1 -> 36.2 | Hurt |
| Adapt with IRG rather than SFOD | Source-free detector adaptation · Sim10K → Cityscapes | AP of Car: 42.3 -> 45.2 | Helped |
| Adapt with IRG rather than SFOD | Source-free detector adaptation · KITTI → Cityscapes | AP of Car: 43.6 -> 46.9 | Helped |
| Adapt with IRG rather than the best listed SFDA baseline | Source-free detector adaptation · PASCAL-VOC → Watercolor | mAP: 47.9 -> 53.0 | Helped |
Closest papers by meaning
- A Free Lunch for Unsupervised Domain Adaptive Object Detection without Source Data (AAAI 2021 · similarity 0.92)
- CAT: Exploiting Inter-Class Dynamics for Domain Adaptive Object Detection (CVPR 2024 · similarity 0.88)
- DA-Ada: Learning Domain-Aware Adapter for Domain Adaptive Object Detection (NeurIPS 2024 · similarity 0.88)
- Unified Domain Generalization and Adaptation for Multi-View 3D Object Detection (NeurIPS 2024 · similarity 0.88)
- Understanding and Improving Source-free Domain Adaptation from a Theoretical Perspective (CVPR 2024 · similarity 0.87)
- Active Domain Adaptation with False Negative Prediction for Object Detection (CVPR 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.