Research map / Detection and segmentation / Open-set and few-shot detection
Periodically Exchange Teacher-Student for Source-Free Object Detection
Qipeng Liu, Luojun Lin, Zhifeng Shen, Zhifeng Yang
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 periodic student/static-teacher weight exchange on top of two-teacher framework (ablation) | PETS · C2F (all levels) | mAP: 36.6 -> 40.3 | Improved |
| add exchange on top of two-teacher framework (ablation) | PETS · K2C | mAP: 43.8 -> 47.0 | Improved |
| use two-way exchange S<->ST instead of one-way S->ST (ablation) | PETS · C2F (all levels) | mAP: 39.6 -> 40.3 | Improved |
| use dynamic teacher to student one-way flow (ablation) | PETS baseline · C2F (all levels) | mAP: 36.6 -> 37.8 | Improved |
| use multi-teacher (dynamic + static + consensus) instead of single dynamic teacher (ablation) | PETS · C2F (all levels) | mAP: 38.0 -> 40.3 | Improved |
| use multi-teacher instead of single dynamic teacher (ablation) | PETS · C2F (single level) | mAP: 32.9 -> 35.9 | Improved |
| use dynamic teacher instead of static teacher as single teacher (ablation) | single-teacher SFOD · C2F (single level) | mAP: 27.2 -> 32.9 | Improved |
| PETS vs best prior SFOD on C2B | PETS · C2B | mAP: 31.3 vs A2SFOD 31.6 | No clear effect |
| PETS applied to S2C | PETS · S2C | mAP: 44.0 -> 57.8 | Improved |
| PETS on train class in C2F | PETS · C2F per-class AP (Train) | AP: Ours 5.5 vs source-only 10.1 | Hurt |
Closest papers by meaning
- Dynamic Retraining-Updating Mean Teacher for Source-Free Object Detection (ECCV 2024 · similarity 0.93)
- Simplifying Source-Free Domain Adaptation for Object Detection: Effective Self-Training Strategies and Performance Insights (ECCV 2024 · similarity 0.92)
- Cross-Domain Adaptive Teacher for Object Detection (CVPR 2022 · similarity 0.90)
- Expert-Teacher-Student Collaborative Learning for Domain Adaptive Object Detection (CVPR 2026 · similarity 0.89)
- Large Self-Supervised Models Bridge the Gap in Domain Adaptive Object Detection (CVPR 2025 · similarity 0.88)
- CAT: Exploiting Inter-Class Dynamics 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.
