Research map / Detection and segmentation / Weakly supervised detection and segmentation
Cross-Domain Weakly-Supervised Object Detection Through Progressive Domain Adaptation
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 |
|---|---|---|---|
| Sequentially fine-tune with domain-transferred images and target-label-guided pseudo-boxes | SSD300 · Clipart1k | mAP: 26.8 -> 46.0 | Helped |
| Fine-tune on domain-transferred source images (ablation) | SSD300 · Clipart1k | mAP: 26.8 -> 38.0 | Helped |
| Fine-tune using target image labels and top-1 pseudo-boxes (ablation) | SSD300 · Clipart1k | mAP: 26.8 -> 36.4 | Helped |
| Replace target image-level labels with one highest-confidence detection per image (ablation) | PL · Clipart1k | mAP: 36.4 -> 25.3 | Hurt |
| Replace target image-level labels with one highest-confidence detection per image during the PL stage (ablation) | DT+PL · Clipart1k | mAP: 46.0 -> 32.7 | Hurt |
| Use DT+PL rather than DT alone (ablation) | SSD300 · Watercolor2k | mAP: 50.4 -> 54.3 | Helped |
| Use DT+PL rather than PL alone (ablation) | SSD300 · Comic2k | mAP: 32.9 -> 37.2 | Helped |
| Add pseudo-labeled extra BAM! images with noisy image-level labels (ablation) | PL · Watercolor2k | mAP: 54.0 -> 59.1 | Helped |
| Add pseudo-labeled extra BAM! images with noisy image-level labels (ablation) | DT+PL · Comic2k | mAP: 37.2 -> 42.2 | Helped |
| Use unweighted detection ensemble rather than the SSD300 baseline | SSD300, CLNet, WSDDN · Clipart1k | mAP: 26.8 -> 26.7 | No clear effect |
| Use feature-distribution alignment with ADDA rather than the SSD300 baseline | SSD300 · Comic2k | mAP: 24.9 -> 23.8 | Hurt |
Closest papers by meaning
- Informative and Consistent Correspondence Mining for Cross-Domain Weakly Supervised Object Detection (CVPR 2021 · similarity 0.92)
- Cross-Domain Adaptive Teacher for Object Detection (CVPR 2022 · similarity 0.91)
- Weakly Supervised Object Localization With Progressive Domain Adaptation (CVPR 2016 · similarity 0.90)
- Adapting Object Detectors via Selective Cross-Domain Alignment (CVPR 2019 · similarity 0.90)
- CAT: Exploiting Inter-Class Dynamics for Domain Adaptive Object Detection (CVPR 2024 · similarity 0.88)
- Universal Domain Adaptive Object Detection via Dual Probabilistic Alignment (AAAI 2025 · 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.