Research map / Label-efficient and robust learning / Domain adaptation and generalization
DLOW: Domain Flow for Adaptation and Generalization
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 |
|---|---|---|---|
| Use DLOW intermediate-domain images instead of target-style-only images | Pixel-level segmentation training on GTA5 · GTA5 → Cityscapes | mIoU: 40.7 to 42.3 | Helped |
| Set z=1 for all translated source images rather than sampling intermediate domainness (ablation) | DLOW pixel-level segmentation training · GTA5 → Cityscapes | mIoU: 42.3 -> 40.7 | Hurt |
| Replace original source images with DLOW-translated intermediate-domain images | AdaptSegNet · GTA5 → Cityscapes | mIoU: 42.4 -> 44.8 | Helped |
| Replace original source images with DLOW-translated intermediate-domain images | AdaptSegNet · GTA5 → KITTI | mIoU: 30.7 -> 36.6 | Helped |
| Replace original source images with DLOW-translated intermediate-domain images | AdaptSegNet · GTA5 → WildDash | mIoU: 18.9 -> 24.9 | Helped |
| Replace original source images with DLOW-translated intermediate-domain images | AdaptSegNet · GTA5 → BDD100K | mIoU: 37.0 -> 39.1 | Helped |
| Use DLOW intermediate-domain images instead of target-style-only images | Pixel-level segmentation training on SYNTHIA · SYNTHIA → Cityscapes | mIoU: 41.6 to 42.8 | Helped |
| Set z=1 for all translated source images rather than sampling intermediate domainness (ablation) | DLOW pixel-level segmentation training · SYNTHIA → Cityscapes | mIoU: 42.8 -> 41.6 | Hurt |
| Replace original source images with DLOW-translated intermediate-domain images | AdaptSegNet · SYNTHIA → Cityscapes | mIoU: 45.7% to 47.1% | Helped |
| Replace original source images with DLOW-translated intermediate-domain images | AdaptSegNet · SYNTHIA → WildDash | mIoU: 20.0 -> 24.4 | Helped |
| Replace original source images with DLOW-translated intermediate-domain images | AdaptSegNet · SYNTHIA → BDD100K | mIoU: 30.8 -> 35.3 | Helped |
| Use DLOW instead of FadNet for photo-to-art style transfer | Photo-to-art image translation · Photo → Van Gogh | user preference: 1.4% / 98.6% | Helped |
Closest papers by meaning
- Deliberated Domain Bridging for Domain Adaptive Semantic Segmentation (NeurIPS 2022 · similarity 0.87)
- Pixel-Level Cycle Association: A New Perspective for Domain Adaptive Semantic Segmentation (NeurIPS 2020 · similarity 0.87)
- Domain-Agnostic Prior for Transfer Semantic Segmentation (CVPR 2022 · similarity 0.87)
- ADAS: A Direct Adaptation Strategy for Multi-Target Domain Adaptive Semantic Segmentation (CVPR 2022 · similarity 0.87)
- Exploring Semantic Consistency and Style Diversity for Domain Generalized Semantic Segmentation (AAAI 2025 · similarity 0.87)
- Consistency Regularization with High-dimensional Non-adversarial Source-guided Perturbation for Unsupervised Domain Adaptation in Segmentation (AAAI 2021 · 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.