Research map / Detection and segmentation / Weakly supervised detection and segmentation
MUM: Mix Image Tiles and UnMix Feature Tiles for Semi-Supervised 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 |
|---|---|---|---|
| add MUM tile mixing and feature unmixing | Unbiased Teacher with Faster R-CNN ResNet-50 · COCO-Standard 0.5% | AP50:95: 16.94±0.23 -> 18.54±0.48 | Helped |
| add MUM tile mixing and feature unmixing | Unbiased Teacher with Faster R-CNN ResNet-50 · COCO-Standard 1% | AP50:95: 20.75±0.12 -> 21.88±0.12 | Helped |
| add MUM tile mixing and feature unmixing | Unbiased Teacher with Faster R-CNN ResNet-50 · COCO-Standard 2%, 5%, 10%, and COCO-Additional | AP50:95: 24.30±0.07 -> 24.84±0.10; 28.27±0.11 -> 28.52±0.09; 31.50±0.10 -> 31.87±0.30; 41.3 -> 42.11 | Helped |
| add MUM tile mixing and feature unmixing | Unbiased Teacher with Faster R-CNN ResNet-50 · VOC07 labeled, VOC12 unlabeled | AP50:95: 48.7 -> 50.22 | Helped |
| add MUM tile mixing and feature unmixing | Unbiased Teacher with Faster R-CNN ResNet-50 · VOC07 labeled, VOC12+COCO20cls unlabeled | AP50:95: 50.34 -> 52.31 | Helped |
| add MUM tile mixing and feature unmixing | Unbiased Teacher with Swin-T · COCO-Standard protocols | AP50:95: 15.95 -> 16.52; 19.8 -> 20.5; 24 -> 24.5; 27.88 -> 28.35; 30.48 -> 30.58 | Helped |
| set tile count per image axis NT to 2 rather than 4 (ablation) | MUM · COCO-Standard 1% ablation | AP50:95: 18.99 -> 18.52 | Hurt |
| set tile count per image axis NT to 8 rather than 4 (ablation) | MUM · COCO-Standard 1% ablation | AP50:95: 18.99 -> 18.28 | Hurt |
| set tile count per image axis NT to 16 rather than 4 (ablation) | MUM · COCO-Standard 1% ablation | AP50:95: 18.99 -> 16.46 | Hurt |
| set group size NG to 2 rather than 4 (ablation) | MUM · COCO-Standard 1% ablation | AP50:95: 18.99 -> 18.92 | No clear effect |
| set group size NG to 6 rather than 4 (ablation) | MUM · COCO-Standard 1% ablation | AP50:95: 18.99 -> 18.85 | No clear effect |
| set group size NG to 12 rather than 4 (ablation) | MUM · COCO-Standard 1% ablation | AP50:95: 18.99 -> 18.84 | No clear effect |
Closest papers by meaning
- Unbiased Teacher for Semi-Supervised Object Detection (ICLR 2021 · similarity 0.88)
- Hierarchical Supervision and Shuffle Data Augmentation for 3D Semi-Supervised Object Detection (CVPR 2023 · similarity 0.87)
- Supervision Interpolation via LossMix: Generalizing Mixup for Object Detection and Beyond (AAAI 2024 · similarity 0.86)
- Semi-supervised 3D Object Detection with PatchTeacher and PillarMix (AAAI 2024 · similarity 0.86)
- Relational Matching for Weakly Semi-Supervised Oriented Object Detection (CVPR 2024 · similarity 0.86)
- Sparse Semi-DETR: Sparse Learnable Queries for Semi-Supervised Object Detection (CVPR 2024 · 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.