Research map / Detection and segmentation / Weakly supervised detection and segmentation
MaskBooster: End-to-End Self-Training for Sparsely Supervised Instance Segmentation
Shida Zheng, Chenshu Chen, Xi Yang, Wenming Tan
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 MaskBooster self-training | Mask RCNN · COCO 0.1% | mask AP: 25.4 -> 31.1 | Improved |
| Add MaskBooster self-training | ShapeProp · COCO 0.1% | mask AP: 28.4 -> 33.3 | Improved |
| Add MaskBooster self-training | Mask RCNN · BDD100K (ρ ≈11%) | mask AP: 27.2 -> 27.9 | Improved |
| Add MaskBooster self-training | ShapeProp · BDD100K (ρ ≈11%) | mask AP: 27.9 -> 29.1 | Improved |
| Add MaskBooster self-training | OPMask · partial-COCO voc → non-voc | AP: 28.5 -> 30.2 | Improved |
| Add MaskBooster self-training | OPMask · partial-COCO non-voc → voc | AP: 36.2 -> 37.6 | Improved |
| Add EMA teacher and strong augmentation (ablation) | Mask RCNN on SpSIS · COCO 0.1% | mask AP: 25.4 -> 26.4 | Improved |
| Add binary pseudo masks refined with box prior (ablation) | EMA+Aug Mask RCNN · COCO 0.1% | mask AP: 26.4 -> 28.0 | Improved |
| Add soft pseudo masks (ablation) | EMA+Aug Mask RCNN · COCO 0.1% | mask AP: 26.4 -> 28.8 | Improved |
| Combine EMA+Aug, binary pseudo masks with box prior, and soft pseudo masks (ablation) | Mask RCNN on SpSIS · COCO 0.1% | mask AP: 25.4 -> 31.1 | Improved |
| Add soft pseudo masks to EMA+Aug and binary pseudo masks (ablation) | Mask RCNN on SpSIS · COCO 1% | mask AP: 33.9 -> 34.1 | Improved |
| Set pseudo-mask loss reweighting length to 1k instead of 20k (ablation) | MaskBooster · COCO 0.1% | mask AP: 31.1 -> 18.1 | Hurt |
Closest papers by meaning
- BoxInst: High-Performance Instance Segmentation With Box Annotations (CVPR 2021 · similarity 0.91)
- Prior to Segment: Foreground Cues for Weakly Annotated Classes in Partially Supervised Instance Segmentation (ICCV 2021 · similarity 0.90)
- BoxTeacher: Exploring High-Quality Pseudo Labels for Weakly Supervised Instance Segmentation (CVPR 2023 · similarity 0.89)
- SOLO: Segmenting Objects by Locations (ECCV 2020 · similarity 0.89)
- Training-Free Dataset Pruning for Instance Segmentation (ICLR 2025 · similarity 0.88)
- MaskDiff: Modeling Mask Distribution with Diffusion Probabilistic Model for Few-Shot Instance Segmentation (AAAI 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.
