Research map / Detection and segmentation / Open-set and few-shot detection
Slim Scissors: Segmenting Thin Object from Synthetic Background
Kunyang Han, Jun Hao Liew, Jiashi Feng, Huawei Tian, Yao Zhao, Yunchao Wei
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 synthetic background input to TPR (ablation) | Slim Scissors · COIFT | IoUthin: 76.1 -> 79.0 | Improved |
| add synthetic background input to TPR (ablation) | Slim Scissors · HRSOD | IoUthin: 60.5 -> 68.6 | Improved |
| add coarse scribbles on top of synthetic background (ablation) | Slim Scissors · COIFT | IoUthin: 79.0 -> 80.8 | Improved |
| add coarse scribbles on top of synthetic background (ablation) | Slim Scissors · HRSOD | IoUthin: 68.6 -> 71.0 | Improved |
| add dual path architecture with TPR instead of single branch with same inputs (ablation) | Slim Scissors · COIFT | IoUthin: 77.9 -> 80.8 | Improved |
| add dual path with TPR versus IOG-like single path (ablation) | Slim Scissors · COIFT | IoUthin: 73.1 -> 80.8 | Improved |
| use thicker scribbles (1.5x radius vs 0.5x) | Slim Scissors · COIFT | IoUthin: 81.4 -> 80.1 | Hurt |
| use thicker scribbles (1.5x radius vs 0.5x) | Slim Scissors · HRSOD | IoUthin: 71.7 -> 69.8 | Hurt |
| add similar detection module (SDM) with thin scribbles | Slim Scissors · COIFT user study | total annotation time (s): 215 -> 72.9 | Improved |
| use thick scribbles instead of thin scribbles | Slim Scissors · COIFT user study (20 images) | IoUthin: 74.8 -> 73.4 | Hurt |
| use Fast Slim Scissors (SDM) instead of full annotation | Slim Scissors · COIFT subset | IoUthin: 81.7 -> 79.9 | Mixed |
Closest papers by meaning
- FocusCut: Diving Into a Focus View in Interactive Segmentation (CVPR 2022 · similarity 0.86)
- GaussianCut: Interactive segmentation via graph cut for 3D Gaussian Splatting (NeurIPS 2024 · similarity 0.85)
- Interactive Boundary Prediction for Object Selection (ECCV 2018 · similarity 0.85)
- AGILE3D: Attention Guided Interactive Multi-object 3D Segmentation (ICLR 2024 · similarity 0.85)
- Extract Free Dense Labels from CLIP (ECCV 2022 · similarity 0.85)
- SLiMe: Segment Like Me (ICLR 2024 · similarity 0.84)
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.
