Research map / Label-efficient and robust learning / Domain adaptation and generalization
Contrastive Mean Teacher for Domain Adaptive Object Detectors
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 object-level contrastive learning to AT | AT · Cityscapes → Foggy Cityscapes (0.02) | mAP: 49.3 -> 50.3 (+1.0) | Helped |
| Add object-level contrastive learning to AT | AT · Cityscapes → Foggy Cityscapes (All) | mAP: 49.8 -> 51.9 (+2.1) | Helped |
| Add object-level contrastive learning to PT | PT · Cityscapes → Foggy Cityscapes (All) | mAP: 47.1 -> 49.3 (+2.2) | Helped |
| Add object-level contrastive learning to PT | PT · KITTI → Cityscapes | AP (Car): 60.2 -> 64.3 (+4.1) | Helped |
| Add object-level contrastive learning to AT | AT · Pascal VOC → Clipart1k | mAP: 45.7 -> 47.0 (+1.3) | Helped |
| Add object-level contrast without class-based contrast or multi-scale features (ablation) | PT · Cityscapes → Foggy Cityscapes (All) | mAP: 47.1 -> 47.8 (+0.7) | Helped |
| Add multi-scale features while omitting class-based contrast (ablation) | PT + object-level contrast · Cityscapes → Foggy Cityscapes (All) | mAP: 47.8 -> 48.2 | Helped |
| Add class-based contrast while omitting multi-scale features (ablation) | PT + object-level contrast · Cityscapes → Foggy Cityscapes (All) | mAP: 47.8 -> 48.7 | Helped |
| Add multi-scale features to class-based object-level contrast (ablation) | PT + CMT · Cityscapes → Foggy Cityscapes (All) | mAP: 48.7 -> 49.3 | Helped |
| Add class-based contrast to multi-scale object-level contrast (ablation) | PT + CMT · Cityscapes → Foggy Cityscapes (All) | mAP: 48.2 -> 49.3 | Helped |
| Use object-level contrastive learning under injected pseudo-label class noise | AT · Cityscapes → Foggy Cityscapes (All), noise level 1.0 | mAP: standard deviation 1.4% -> 0.4%; gain +2.0% -> +2.8% | Helped |
Closest papers by meaning
- Cross-Domain Adaptive Teacher for Object Detection (CVPR 2022 · similarity 0.92)
- Unbiased Mean Teacher for Cross-Domain Object Detection (CVPR 2021 · similarity 0.90)
- CAT: Exploiting Inter-Class Dynamics for Domain Adaptive Object Detection (CVPR 2024 · similarity 0.89)
- Large Self-Supervised Models Bridge the Gap in Domain Adaptive Object Detection (CVPR 2025 · similarity 0.89)
- Mean Teacher DETR with Masked Feature Alignment: A Robust Domain Adaptive Detection Transformer Framework (AAAI 2024 · similarity 0.88)
- D3T: Distinctive Dual-Domain Teacher Zigzagging Across RGB-Thermal Gap for Domain-Adaptive 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.