Research map / Detection and segmentation / Open-set and few-shot detection
Positive-Unlabeled Data Purification in the Wild for Object Detection
Check 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 |
|---|---|---|---|
| Use PU classification to select unlabeled images instead of random selection | FPN object detection on COCO · COCO minival | mAP: 38.2 -> 39.4 | Helped |
| Add PU-selected unlabeled ImageNet images to labeled training data | FPN · COCO minival | mAP: 37.4 -> 39.4 | Helped |
| Increase randomly selected unlabeled data from 118k to 1.3M images | FPN · COCO minival | mAP: 37.9 -> 37.0 | Hurt |
| Add self-distillation to a ResNet-50 FPN detector (ablation) | FPN · COCO minival | mAP: 37.4 -> 38.4 | Helped |
| Add multi-layer hint learning (ablation) | ResNet-18 FPN student with ResNet-50 FPN teacher · COCO minival | mAP: 33.4 -> 34.6 | Helped |
| Add ground-truth-bounded classification distillation (ablation) | ResNet-18 FPN student with ResNet-50 FPN teacher · COCO minival | mAP: 34.6 -> 35.3 | Helped |
| Add ground-truth-bounded regression distillation (ablation) | ResNet-18 FPN student with ResNet-50 FPN teacher · COCO minival | mAP: 35.3 -> 35.8 | Helped |
| Add multi-layer hint learning (ablation) | ResNet-50 FPN teacher and student · COCO minival | mAP: 37.4 -> 37.8 | Helped |
| Add ground-truth-bounded classification distillation (ablation) | ResNet-50 FPN teacher and student · COCO minival | mAP: 37.8 -> 38.2 | Helped |
| Add ground-truth-bounded regression distillation (ablation) | ResNet-50 FPN teacher and student · COCO minival | mAP: 38.2 -> 38.4 | Helped |
| Lower the pseudo-label threshold from 0.9 to 0.5 for PU-selected data | FPN with 118k PU-selected images · COCO minival | mAP: 39.0 -> 39.4 | Helped |
Closest papers by meaning
- Bridging Non Co-occurrence with Unlabeled In-the-wild Data for Incremental Object Detection (NeurIPS 2021 · similarity 0.86)
- Unbiased Teacher for Semi-Supervised Object Detection (ICLR 2021 · similarity 0.85)
- A Free Lunch for Unsupervised Domain Adaptive Object Detection without Source Data (AAAI 2021 · similarity 0.85)
- Towards Human-Machine Cooperation: Self-Supervised Sample Mining for Object Detection (CVPR 2018 · similarity 0.85)
- OD$^3$: Optimization-free Dataset Distillation for Object Detection (ICLR 2026 · similarity 0.85)
- SimLTD: Simple Supervised and Semi-Supervised Long-Tailed Object Detection (CVPR 2025 · similarity 0.85)
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.