Research map / Detection and segmentation / Weakly supervised detection and segmentation
Calibrate Before Adapt: Training-Free Pseudo-Label Calibration for Semi-Supervised Cross-Domain Few-Shot Detection
Xin Yang, Fei Zhou, Wei Wei, Lei 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 prototype-based semantic calibrator (ablation) | MM-Grounding-DINO baseline · DIOR 1-shot | mAP: 18.6 -> 20.6 | Improved |
| add prototype-based semantic calibrator (ablation) | MM-Grounding-DINO baseline · DIOR 5-shot | mAP: 29.6 -> 32.1 | Improved |
| add prototype-based semantic calibrator (ablation) | MM-Grounding-DINO baseline · DIOR 10-shot | mAP: 35.7 -> 36.0 | Improved |
| add prototype-based semantic calibrator (ablation) | MM-Grounding-DINO baseline · NEU-DET 1-shot | mAP: 13.9 -> 16.5 | Improved |
| add prototype-based semantic calibrator (ablation) | MM-Grounding-DINO baseline · NEU-DET 5-shot | mAP: 22.2 -> 23.5 | Improved |
| add prototype-based semantic calibrator (ablation) | MM-Grounding-DINO baseline · NEU-DET 10-shot | mAP: 24.3 -> 25.6 | Improved |
| add visual self-prompting with SAM on top of calibrator (ablation) | calibrated pseudo-labels · DIOR 1-shot | mAP: 20.6 -> 23.7 | Improved |
| add visual self-prompting with SAM on top of calibrator (ablation) | calibrated pseudo-labels · DIOR 5-shot | mAP: 32.1 -> 32.5 | Improved |
| add visual self-prompting with SAM on top of calibrator (ablation) | calibrated pseudo-labels · NEU-DET 5-shot | mAP: 23.5 -> 23.8 | Improved |
| pseudo-label filtering with threshold 0 instead of calibration | MM-Grounding-DINO · six-dataset average 1-shot | mAP: 31.2 -> 19.7 | Hurt |
| use our calibration framework instead of threshold filtering thr=0.9 | MM-Grounding-DINO pseudo-labels · six-dataset average 1-shot | mAP: 32.8 -> 35.9 | Improved |
| use visual prompts instead of text prompts for SAM3 | SAM3 · six-dataset average 1-shot | mAP: 28.8 -> 32.1 | Improved |
Closest papers by meaning
- Label, Verify, Correct: A Simple Few Shot Object Detection Method (CVPR 2022 · similarity 0.90)
- Mind the Gap: Polishing Pseudo Labels for Accurate Semi-supervised Object Detection (AAAI 2023 · similarity 0.90)
- Adapting Object Size Variance and Class Imbalance for Semi-supervised Object Detection (AAAI 2023 · similarity 0.90)
- Rethinking Pseudo Labels for Semi-supervised Object Detection (AAAI 2022 · similarity 0.90)
- Semi-DETR: Semi-Supervised Object Detection With Detection Transformers (CVPR 2023 · similarity 0.89)
- Diverse Learner: Exploring Diverse Supervision for Semi-Supervised Object Detection (ECCV 2022 · similarity 0.89)
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.
