Research map / Label-efficient and robust learning / Domain adaptation and generalization
Style-Hallucinated Dual Consistency Learning for Domain Generalized Semantic Segmentation
Yuyang Zhao, Zhun Zhong, Na Zhao, Nicu Sebe, Gim Hee Lee
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 style hallucination module (ablation) | DeepLabV3+ baseline · GTAV to real, ResNet-50 (mean of 3) | mIoU: 27.42 -> 35.68 | Improved |
| add style consistency (logit pairing with JSD) (ablation) | SHM model · GTAV to real, ResNet-50 (mean of 3) | mIoU: 35.68 -> 39.67 | Improved |
| add retrospection consistency (ablation) | SHM model · GTAV to real, ResNet-50 (mean of 3) | mIoU: 35.68 -> 40.29 | Improved |
| add retrospection consistency (ablation) | SHM + SC model · GTAV to real, ResNet-50 (mean of 3) | mIoU: 39.67 -> 42.42 | Improved |
| replace ImageNet pre-trained model in RC with EMA model (ablation) | SHM + SC model · GTAV to real, ResNet-50 (mean of 3) | mIoU: 42.42 -> 40.92 | Hurt |
| use FPS basis styles instead of Kmeans basis styles (ablation) | SHM · GTAV to real, ResNet-50 (mean of 3) | mIoU: 39.19 -> 42.42 | Improved |
| use SHM instead of MixStyle (ablation) | SHADE · GTAV to real, ResNet-50 (mean of 3) | mIoU: 41.10 -> 42.42 | Improved |
| use SHM instead of CrossNorm (ablation) | SHADE · GTAV to real, ResNet-50 (mean of 3) | mIoU: 40.72 -> 42.42 | Improved |
| use SHM instead of random normal styles (ablation) | SHADE · GTAV to real, ResNet-50 (mean of 3) | mIoU: 37.89 -> 42.42 | Improved |
| add retrospection consistency (ablation) | SHADE · CityScapes per-class IoU (things classes) | mIoU: +2.47% | Improved |
| insert SHM after layer3 instead of layer0 (ablation) | SHADE · GTAV to real, ResNet-50 (mean of 3) | mIoU | Hurt |
| select basis styles only once instead of every 3 epochs (ablation) | SHM · GTAV to real, ResNet-50 (mean of 3) | mIoU: 42.42 -> lower than 41 | Hurt |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- FSDR — FSDR: Frequency Space Domain Randomization for Domain Generalization (CVPR 2021) · 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
- Style-Hallucinated Dual Consistency Learning: A Unified Framework for Visual Domain Generalization (IJCV 2023 · similarity 0.94)
- Exploring Semantic Consistency and Style Diversity for Domain Generalized Semantic Segmentation (AAAI 2025 · similarity 0.92)
- Style Blind Domain Generalized Semantic Segmentation via Covariance Alignment and Semantic Consistence Contrastive Learning (CVPR 2024 · similarity 0.90)
- Semantic-Aware Domain Generalized Segmentation (CVPR 2022 · similarity 0.89)
- Do We Need Perfect Data? Leveraging Noise for Domain Generalized Segmentation (AAAI 2026 · similarity 0.89)
- Collaborating Foundation Models for Domain Generalized Semantic Segmentation (CVPR 2024 · similarity 0.87)
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.
