Research map / Label-efficient and robust learning / Domain adaptation and generalization
Adversarial Style Augmentation for Domain Generalized Urban-Scene Segmentation
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 AdvStyle adversarial style augmentation | DeepLabV3+ ResNet-50 baseline · GTAV to real mean | mIoU: 27.42 -> 37.39 | Helped |
| add AdvStyle | IBN-Net ResNet-101 · GTAV to real mean | mIoU: 36.13 -> 42.22 | Helped |
| add AdvStyle | ISW ResNet-101 · SYNTHIA to real mean | mIoU: 28.51 -> 33.65 | Helped |
| learn adversarial style instead of pixel-wise adversarial perturbation (AdvPixel) (ablation) | AdvStyle · GTAV to CityScapes/BDD/Mapillary mean, ResNet-50 | mIoU: 33.97 -> 37.39 | Helped |
| replace random style noise with adversarial style learning (ablation) | AdvStyle · GTAV to CityScapes/BDD/Mapillary mean, ResNet-50 | mIoU: 33.07 -> 37.39 | Helped |
| replace MixStyle or CrossStyle with adversarial style (ablation) | AdvStyle · GTAV to CityScapes/BDD/Mapillary mean, ResNet-50 | mIoU: 34.60 -> 37.39 | Helped |
| add color jittering and Gaussian blur to AdvStyle (ablation) | AdvStyle · GTAV to CityScapes/BDD/Mapillary mean, ResNet-50 | mIoU: 35.32 -> 37.39 | Helped |
| add AdvPixel to AdvStyle (ablation) | AdvStyle · GTAV to CityScapes/BDD/Mapillary mean, ResNet-50 | mIoU: 37.39 -> 38.20 | Helped |
| inject AdvStyle at image level instead of deep layers (ablation) | AdvStyle · GTAV to CityScapes/BDD/Mapillary mean, ResNet-50 | mIoU: 27.44 -> 37.39 | Helped |
| use per-patch style features instead of whole-image style (ablation) | AdvStyle · GTAV to CityScapes/BDD/Mapillary mean, ResNet-50 | mIoU: 37.39 -> 37.76 | No clear effect |
| use RGB instead of LAB color space for AdvStyle (ablation) | AdvStyle · GTAV to CityScapes/BDD/Mapillary mean, ResNet-50 | mIoU: 35.70 -> 37.39 | Helped |
| add AdvStyle to ERM | ERM ResNet18 · PACS average | accuracy: 58.9 -> 65.4 | Helped |
Closest papers by meaning
- Style Blind Domain Generalized Semantic Segmentation via Covariance Alignment and Semantic Consistence Contrastive Learning (CVPR 2024 · similarity 0.88)
- Exploring Semantic Consistency and Style Diversity for Domain Generalized Semantic Segmentation (AAAI 2025 · similarity 0.87)
- Learning Content-Enhanced Mask Transformer for Domain Generalized Urban-Scene Segmentation (AAAI 2024 · similarity 0.86)
- MRFP: Learning Generalizable Semantic Segmentation from Sim-2-Real with Multi-Resolution Feature Perturbation (CVPR 2024 · similarity 0.86)
- A Simple Recipe for Language-guided Domain Generalized Segmentation (CVPR 2024 · similarity 0.86)
- Stronger Fewer & Superior: Harnessing Vision Foundation Models for Domain Generalized Semantic Segmentation (CVPR 2024 · 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.