Research map / Detection and segmentation / Weakly supervised detection and segmentation
Pointly-Supervised Panoptic Segmentation
Junsong Fan, Zhaoxiang Zhang, Tieniu 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 boundary criterion to semantic similarity in traversing cost (ablation) | point-supervised panoptic model · VOC val | PQ: 46.6 -> 48.5 | Improved |
| add manifold criterion to semantic similarity in traversing cost (ablation) | point-supervised panoptic model · VOC val | PQ: 46.6 -> 49.4 | Improved |
| add both boundary and manifold criteria (ablation) | point-supervised panoptic model · VOC val | PQ: 46.6 -> 49.8 | Improved |
| add boundary and manifold criteria, effect on stuff classes (ablation) | point-supervised panoptic model · VOC val | PQ-stuff: 89.1 -> 89.5 | No clear effect |
| raise boundary weight lambda_b from 0.1 to 0.5 or 1.0 (ablation) | traversing cost · VOC val | PQ: 49.8 -> 48.3 | Hurt |
| raise manifold weight lambda_m from 0.1 to 1.0 (ablation) | traversing cost · VOC val | PQ: 49.8 -> 48.9 | Hurt |
| sample points near object centers instead of uniformly (ablation) | point-supervised panoptic model · VOC val | PQ: 49.8 -> 50.9 | Improved |
| sample points near borders instead of uniformly (ablation) | point-supervised panoptic model · VOC val | PQ: 49.8 -> 44.7 | Hurt |
| add color prior loss to the semantic segmentation branch (ablation) | point-supervised panoptic model · VOC val | mIoU: 62.2 -> 67.5 | Improved |
| add color prior loss, effect on panoptic quality (ablation) | point-supervised panoptic model · VOC val | PQ: 41.3 -> 49.8 | Improved |
| use 10 points per target instead of 1 | point-supervised panoptic model · COCO val | PQ: 29.3 -> 33.1 | Improved |
| use 10 points per target instead of 1 | point-supervised panoptic model · VOC val | PQ: 49.8 -> 56.6 | Improved |
Closest papers by meaning
- Fully Data-Driven Pseudo Label Estimation for Pointly-Supervised Panoptic Segmentation (AAAI 2024 · similarity 0.94)
- Point-supervised Panoptic Segmentation via Estimating Pseudo Labels from Learnable Distance (ECCV 2024 · similarity 0.94)
- Point2Mask: Point-supervised Panoptic Segmentation via Optimal Transport (ICCV 2023 · similarity 0.93)
- Weakly- and Semi-Supervised Panoptic Segmentation (ECCV 2018 · similarity 0.92)
- Fully Convolutional Networks for Panoptic Segmentation with Point-based Supervision (TPAMI 2022 · similarity 0.91)
- Scene-Centric Unsupervised Panoptic Segmentation (CVPR 2025 · similarity 0.89)
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.
