Research map / Label-efficient and robust learning / Domain adaptation and generalization
RobustNet: Improving Domain Generalization in Urban-Scene Segmentation via Instance Selective Whitening
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 |
|---|---|---|---|
| Apply instance selective whitening loss | ResNet-50 segmentation trained on GTAV · BDD-100K | mIoU: 25.14 -> 35.20 | Helped |
| Apply instance selective whitening loss | ResNet-50 segmentation trained on GTAV · Mapillary | mIoU: 28.18 -> 40.33 | Helped |
| Apply instance selective whitening loss | ResNet-50 segmentation trained on GTAV · SYNTHIA | mIoU: 26.23 -> 28.30 | Helped |
| Apply instance selective whitening loss | ResNet-50 segmentation trained on GTAV · GTAV | mIoU: 73.45 -> 72.10 | Hurt |
| Apply instance selective whitening loss | ResNet-50 segmentation trained on Cityscapes · BDD-100K | mIoU: 44.96 -> 50.73 | Helped |
| Apply instance selective whitening loss | ResNet-50 segmentation trained on Cityscapes · Mapillary | mIoU: 51.68 -> 58.64 | Helped |
| Apply instance selective whitening loss | ResNet-50 segmentation trained on Cityscapes · SYNTHIA | mIoU: 23.29 -> 26.20 | Helped |
| Apply instance selective whitening loss | ResNet-50 segmentation trained on Cityscapes · Cityscapes | mIoU: 77.51 -> 76.41 | Hurt |
| Selectively whiten style-sensitive covariances rather than whiten all covariances (ablation) | GTAV-trained segmentation · BDD-100K | mIoU: 32.67 -> 35.20 | Helped |
| Selectively whiten style-sensitive covariances rather than use margin-relaxed whitening (ablation) | GTAV-trained segmentation · Mapillary | mIoU: 38.42 -> 40.33 | No clear effect |
| Set covariance clustering parameter k to 3 rather than 2, 5, 10, or 20 (ablation) | ISW on GTAV-trained segmentation · Cityscapes | mIoU: 35.46; 34.84; 33.58; 33.66 -> 36.58 | Helped |
| Set ISW loss weight gamma to 0.6 rather than 0.4 or 0.8 (ablation) | ISW on GTAV-trained segmentation · BDD-100K | mIoU: 34.07; 34.01 -> 35.20 | Helped |
Closest papers by meaning
- Stronger Fewer & Superior: Harnessing Vision Foundation Models for Domain Generalized Semantic Segmentation (CVPR 2024 · similarity 0.88)
- Learning Content-Enhanced Mask Transformer for Domain Generalized Urban-Scene Segmentation (AAAI 2024 · similarity 0.88)
- Style Blind Domain Generalized Semantic Segmentation via Covariance Alignment and Semantic Consistence Contrastive Learning (CVPR 2024 · similarity 0.86)
- MRFP: Learning Generalizable Semantic Segmentation from Sim-2-Real with Multi-Resolution Feature Perturbation (CVPR 2024 · similarity 0.86)
- Learning Generalized Segmentation for Foggy-Scenes by Bi-directional Wavelet Guidance (AAAI 2024 · similarity 0.86)
- A Simple Recipe for Language-guided Domain Generalized 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.