Research map / Detection and segmentation / Open-set and few-shot detection
MTTrans: Cross-Domain Object Detection with Mean Teacher Transformer
Jinze Yu, Jiaming Liu, Xiaobao Wei, Haoyi Zhou, Yohei Nakata, Denis Gudovskiy, Tomoyuki Okuno, Jianxin Li, Kurt Keutzer, Shanghang Zhang
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 mean teacher framework with pseudo labels (ablation) | Deformable DETR · Cityscapes to Foggy Cityscapes | mAP50: 28.5 -> 35.843 (+7.34) | Improved |
| add multi-level feature alignment on top of mean teacher (ablation) | MT-DefDETR · Cityscapes to Foggy Cityscapes | mAP50: 35.843 -> 43.413 | Improved |
| add DQFA on encoder (ablation) | MTTrans · Cityscapes to Foggy Cityscapes | mAP50: 43.183 -> 43.413 | Improved |
| add DQFA on decoder (ablation) | MTTrans · Cityscapes to Foggy Cityscapes | mAP50: 42.962 -> 43.413 | Improved |
| add TIFA on encoder (ablation) | MTTrans · Cityscapes to Foggy Cityscapes | mAP50: 43.013 -> 43.413 | Improved |
| add BGPA on decoder (ablation) | MTTrans · Cityscapes to Foggy Cityscapes | mAP50: 43.003 -> 43.413 | Improved |
| use BGPA instead of TIFA for decoder alignment (ablation) | MTTrans · Cityscapes to Foggy Cityscapes | mAP50: 42.953 -> 43.413 | Improved |
| use TIFA instead of BGPA for encoder alignment (ablation) | MTTrans · Cityscapes to Foggy Cityscapes | mAP50: 43.068 -> 43.413 | Improved |
| share object queries between teacher and student (ablation) | MTTrans · Cityscapes to Foggy Cityscapes | mAP50: 42.797 -> 43.413 | Improved |
| MTTrans versus SFA on synthetic to real | Deformable DETR · Sim10k to Cityscapes | car AP: 52.6 -> 57.9 | Improved |
| MTTrans versus SFA on scene adaptation | Deformable DETR · Cityscapes to BDD100k daytime | mAP50: 28.9 -> 32.6 | Improved |
| MTTrans versus ViSGA on weather adaptation | MTTrans · Cityscapes to Foggy Cityscapes | mAP50: 43.3 -> 43.4 | No clear effect |
Later papers that compared with this method
4 later accepted papers report MTTrans as a baseline in their results tables; 3 of them report a better number on most of the results they share with it.
- Beyond Boundaries: Leveraging Vision Foundation Models for Source-Free Object Detection (AAAI 2026 · better on 1 of 1 shared result)
- DA-BEV: Unsupervised Domain Adaptation for Bird's Eye View Perception (ECCV 2024 · better on 3 of 3 shared results)
- Masked Retraining Teacher-Student Framework for Domain Adaptive Object Detection (ICCV 2023 · better on 3 of 3 shared results)
- DA-DETR: Domain Adaptive Detection Transformer With Information Fusion (CVPR 2023 · better on 1 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
- Mean Teacher DETR with Masked Feature Alignment: A Robust Domain Adaptive Detection Transformer Framework (AAAI 2024 · similarity 0.93)
- Unbiased Mean Teacher for Cross-Domain Object Detection (CVPR 2021 · similarity 0.93)
- Exploring Object Relation in Mean Teacher for Cross-Domain Detection (CVPR 2019 · similarity 0.92)
- Cross-Domain Adaptive Teacher for Object Detection (CVPR 2022 · similarity 0.91)
- Adapting Object Detectors via Selective Cross-Domain Alignment (CVPR 2019 · similarity 0.91)
- Harmonious Teacher for Cross-Domain Object Detection (CVPR 2023 · similarity 0.90)
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.
