Research map / Detection and segmentation / Weakly supervised detection and segmentation
Box-Supervised Instance Segmentation with Level Set Evolution
Wentong Li, Wenyu Liu, Jianke Zhu, Miaomiao Cui, Xian-Sheng Hua, Lei Zhang
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 image-based level set energy to box projection (ablation) | Level set instance seg · Pascal VOC val | AP: 19.7 -> 22.2 | Improved |
| add high-level feature energy term (ablation) | Level set instance seg · Pascal VOC val | AP: 22.2 -> 24.7 | Improved |
| restrict level set evolution to the box region (ablation) | Level set instance seg · Pascal VOC val | AP: 21.7 -> 24.7 | Improved |
| use 9 feature channels instead of 10 (ablation) | Level set instance seg · Pascal VOC val | AP: 22.0 -> 24.7 | Improved |
| use 9 feature channels instead of 5 (ablation) | Level set instance seg · Pascal VOC val | AP: 23.3 -> 24.7 | Improved |
| train 3x multi-scale instead of 1x (ablation) | Level set instance seg · Pascal VOC val | AP: 24.7 -> 34.4 | Improved |
| add tree filter on deep features (ablation) | Level set instance seg · Pascal VOC val | AP: 34.4 -> 36.3 | Improved |
| level set method vs BoxInst | Ours · iSAID val | AP: 17.8 -> 20.1 | Improved |
| level set method vs BoxInst | Ours · LiTS val | AP: 40.7 -> 44.5 | Improved |
| level set method vs BoxInst on small objects | Ours (ResNet-101-DCN) · COCO test-dev | AP_S: 17.1 -> 16.8 | Hurt |
| level set method vs BoxInst | Ours (ResNet-101) · COCO test-dev | AP: 33.2 -> 33.4 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 3 of 3 on most shared results.
- BoxInst — BoxInst: High-Performance Instance Segmentation With Box Annotations (CVPR 2021) · better on 10 of 11 shared results
- BBAM — BBAM: Bounding Box Attribution Map for Weakly Supervised Semantic and Instance Segmentation (CVPR 2021) · better on 3 of 3 shared results
- DiscoBox — DiscoBox: Weakly Supervised Instance Segmentation and Semantic Correspondence From Box Supervision (ICCV 2021) · better on 2 of 2 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Box2Mask: Box-Supervised Instance Segmentation via Level-Set Evolution (TPAMI 2024 · similarity 0.96)
- BoxInst: High-Performance Instance Segmentation With Box Annotations (CVPR 2021 · similarity 0.91)
- SIM: Semantic-Aware Instance Mask Generation for Box-Supervised Instance Segmentation (CVPR 2023 · similarity 0.90)
- BoxTeacher: Exploring High-Quality Pseudo Labels for Weakly Supervised Instance Segmentation (CVPR 2023 · similarity 0.90)
- Learning to Segment Every Thing (CVPR 2018 · similarity 0.90)
- Extreme Point Supervised Instance Segmentation (CVPR 2024 · similarity 0.88)
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.
