Research map / Detection and segmentation / CNN features
UniDAformer: Unified Domain Adaptive Panoptic Segmentation Transformer via Hierarchical Mask Calibration
Read the paperCheck 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 |
|---|---|---|---|
| Add online self-training on unlabeled target images (ablation) | DETR panoptic baseline · SYNTHIA → Cityscapes | mPQ: 18.3 -> 22.6 | Helped |
| Add region-wise mask calibration to self-training (ablation) | DETR panoptic model · SYNTHIA → Cityscapes | mPQ: 22.6 -> 28.7 | Helped |
| Add superpixel-wise mask calibration to self-training (ablation) | DETR panoptic model · SYNTHIA → Cityscapes | mPQ: 22.6 -> 26.2 | Helped |
| Add pixel-wise mask calibration to self-training (ablation) | DETR panoptic model · SYNTHIA → Cityscapes | mPQ: 22.6 -> 24.2 | Helped |
| Add region-wise, superpixel-wise, and pixel-wise calibration (ablation) | DETR panoptic model with self-training · SYNTHIA → Cityscapes | mPQ: 22.6 -> 33.0 | Helped |
| Use coarse-to-fine region → superpixel → pixel calibration instead of reversed orders (ablation) | UniDAformer HMC · SYNTHIA → Cityscapes | mPQ: 29.9; 30.1; 33.0 | Helped |
| Use unified HMC-based adaptation rather than CVRN | DETR panoptic architecture · SYNTHIA → Cityscapes | mPQ: 27.9 -> 33.0 | Helped |
| Use unified HMC-based adaptation rather than CVRN | DETR panoptic architecture · Cityscapes → Foggy Cityscapes | mPQ: 35.7 -> 37.6 | Helped |
| Use unified HMC-based adaptation rather than CVRN | DETR panoptic architecture · VIPER → Cityscapes | mPQ: 31.1 -> 34.5 | Helped |
| Replace the multi-branch CVRN architecture with unified UniDAformer | Panoptic domain adaptation · SYNTHIA → Cityscapes | inference speed: 0.36 fps -> 5.23 fps | Helped |
Closest papers by meaning
- OneFormer: One Transformer To Rule Universal Image Segmentation (CVPR 2023 · similarity 0.90)
- Ada-Segment: Automated Multi-loss Adaptation for Panoptic Segmentation (AAAI 2021 · similarity 0.89)
- Universal Domain Adaptation for Semantic Segmentation (CVPR 2025 · similarity 0.88)
- OneFormer3D: One Transformer for Unified Point Cloud Segmentation (CVPR 2024 · similarity 0.86)
- Unsupervised Universal Image Segmentation (CVPR 2024 · similarity 0.86)
- Towards Unsupervised Domain Bridging via Image Degradation in Semantic Segmentation (NeurIPS 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.