Research map / Detection and segmentation / CNN features
Attention-Guided Unified Network for Panoptic Segmentation
Check your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| Unify foreground and background segmentation with shared features and joint optimization (ablation) | AUNet · MS-COCO val | PQ: 37.2 -> 38.3; 1.1% absolute improvements | Helped |
| Add proposal attention using RPN features (ablation) | AUNet · MS-COCO val | PQ: 38.3 -> 39 | Helped |
| Add mask attention using foreground masks (ablation) | AUNet · MS-COCO val | PQ: 38.3 -> 38.9 | Helped |
| Add background reweighting to PAM (ablation) | PAM · MS-COCO val | PQ: 39 -> 39.4 | Helped |
| Add background reweighting to MAM (ablation) | MAM · MS-COCO val | PQ: 38.9 -> 39.2 | Helped |
| Use both PAM and MAM with background reweighting (ablation) | AUNet · MS-COCO val | PQ: 37.2 -> 39.6; 2.4% | Helped |
| Use 28 × 28 rather than 14 × 14 masks for RoIUpsample (ablation) | AUNet MAM · MS-COCO val | PQ: 0.1% absolute improvement | Helped |
| Treat all things as one category in the background branch instead of using all 19 semantic labels (ablation) | AUNet · Cityscapes val | PQ: 55.0 -> 56.4 | Hurt |
| Disable MAM (ablation) | AUNet · Cityscapes val | PQ | Helped |
| Use ResNeXt-152-FPN instead of ResNet-101-FPN | AUNet · MS-COCO test-dev | PQ: 45.2 -> 46.5 | Helped |
Closest papers by meaning
- Fully Convolutional Networks for Panoptic Segmentation (CVPR 2021 · similarity 0.90)
- Auto-Panoptic: Cooperative Multi-Component Architecture Search for Panoptic Segmentation (NeurIPS 2020 · similarity 0.90)
- Bidirectional Graph Reasoning Network for Panoptic Segmentation (CVPR 2020 · similarity 0.90)
- Unifying Training and Inference for Panoptic Segmentation (CVPR 2020 · similarity 0.90)
- FOCUS: Towards Universal Foreground Segmentation (AAAI 2025 · similarity 0.89)
- Scene-Centric Unsupervised Panoptic Segmentation (CVPR 2025 · similarity 0.86)
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.