Research map / Detection and segmentation / Weakly supervised detection and segmentation
DiscoBox: Weakly Supervised Instance Segmentation and Semantic Correspondence From Box Supervision
Shiyi Lan, Zhiding Yu, Christopher Choy, Subhashree Radhakrishnan, Guilin Liu, Yuke Zhu, Larry S. Davis, Anima Anandkumar
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 self-consistency loss L_con to the teacher (ablation) | DiscoBox YOLACT++ ResNet-50-DCN · VOC12 val | AP50: 43.3 -> 62.0 | Improved |
| add self-consistency loss L_con to the teacher (ablation) | DiscoBox SOLOv2 ResNet-50 · VOC12 val | AP50: 42.1 -> 58.1 | Improved |
| add dense contrastive loss L_nce (ablation) | DiscoBox YOLACT++ ResNet-50-DCN · VOC12 val | AP50: 62.0 -> 63.6 | Improved |
| add dense contrastive loss L_nce (ablation) | DiscoBox SOLOv2 ResNet-50 · VOC12 val | AP50: 58.1 -> 59.8 | Improved |
| add dense contrastive loss L_nce (ablation) | DiscoBox YOLACT++ ResNet-50-DCN · VOC12 val | AP75: 33.5 -> 34.1 | Improved |
| add dense NCE loss during training (teacher still used at inference) (ablation) | DiscoBox · PASCAL 3D+ | AP: 30.9 -> 31.7 | Improved |
| use structured teacher for correspondence instead of SCOT matching (ablation) | DiscoBox- · PASCAL 3D+ | AP: 29.3 -> 30.9 | Improved |
| scale backbone from ResNet-50 to ResNet-101-DCN | DiscoBox SOLOv2 · COCO test-dev | mask AP: 32.0 -> 35.8 | Improved |
| scale backbone from ResNet-101-DCN to ResNeXt-101-DCN | DiscoBox SOLOv2 · COCO test-dev | mask AP: 35.8 -> 37.9 | Improved |
| use SOLOv2 instead of YOLACT++ as task network (ResNet-50) | DiscoBox · VOC12 val | AP50: 63.6 -> 59.8 | Hurt |
| vary loss weights (ablation) | DiscoBox · VOC12 and PASCAL 3D+ | AP | No clear effect |
Later papers that compared with this method
5 later accepted papers report DiscoBox as a baseline in their results tables; 5 of them report a better number on most of the results they share with it.
- Rethinking Box Supervision: Bias-Free Weakly Supervised Medical Segmentation (CVPR 2026 · better on 2 of 2 shared results)
- Box2Mask: Box-Supervised Instance Segmentation via Level-Set Evolution (TPAMI 2024 · better on 5 of 5 shared results)
- BoxTeacher: Exploring High-Quality Pseudo Labels for Weakly Supervised Instance Segmentation (CVPR 2023 · better on 1 of 1 shared result)
- BoxSnake: Polygonal Instance Segmentation with Box Supervision (ICCV 2023 · better on 1 of 1 shared result)
- Box-Supervised Instance Segmentation with Level Set Evolution (ECCV 2022 · better on 2 of 2 shared results)
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 on most shared results.
- BoxInst — BoxInst: High-Performance Instance Segmentation With Box Annotations (CVPR 2021) · better on 3 of 3 shared results
- SF-Net — SFNet: Learning Object-Aware Semantic Correspondence (CVPR 2019) · better on 1 of 1 shared result
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
- BoxTeacher: Exploring High-Quality Pseudo Labels for Weakly Supervised Instance Segmentation (CVPR 2023 · similarity 0.91)
- Simple Does It: Weakly Supervised Instance and Semantic Segmentation (CVPR 2017 · similarity 0.91)
- BoxInst: High-Performance Instance Segmentation With Box Annotations (CVPR 2021 · similarity 0.90)
- SIM: Semantic-Aware Instance Mask Generation for Box-Supervised Instance Segmentation (CVPR 2023 · similarity 0.90)
- WISH: Weakly Supervised Instance Segmentation using Heterogeneous Labels (CVPR 2025 · similarity 0.89)
- Sketchy Bounding-box Supervision for 3D Instance Segmentation (CVPR 2025 · 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.
