Research map / Label-efficient and robust learning / Domain adaptation and generalization
Domain Randomization and Pyramid Consistency: Simulation-to-Real Generalization Without Accessing Target Domain Data
Xiangyu Yue, Yang Zhang, Sicheng Zhao, Alberto Sangiovanni-Vincentelli, Kurt Keutzer, Boqing Gong
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 domain randomization with auxiliary stylized domains (ablation) | FCN8s-VGG16 · GTA -> Cityscapes | mIoU: 30.04 -> 34.64 | Improved |
| add domain randomization (ablation) | FCN8s-VGG16 · GTA -> BDDS | mIoU: 24.59 -> 30.14 | Improved |
| add domain randomization (ablation) | FCN8s-VGG16 · GTA -> Mapillary | mIoU: 26.63 -> 31.64 | Improved |
| add pyramid consistency across domains on top of DR (ablation) | FCN8s-VGG16 · GTA -> Cityscapes | mIoU: 34.64 -> 35.47 | Improved |
| add pyramid consistency within an image on top of DR (ablation) | FCN8s-VGG16 · GTA -> Cityscapes | mIoU: 34.64 -> 35.12 | Improved |
| add both consistency losses on top of DR (ablation) | FCN8s-VGG16 · GTA -> Cityscapes | mIoU: 34.64 -> 36.11 | Improved |
| add both consistency losses on top of DR (ablation) | FCN8s-VGG16 · GTA -> Mapillary | mIoU: 31.64 -> 32.25 | Improved |
| increase number of auxiliary domains (ablation) | FCN8s-VGG16 · GTA -> Cityscapes/BDDS/Mapillary | mIoU: accuracy increases then saturates (Figure 4, no numbers) | Mixed |
| apply full method with ResNet-101 backbone | FCN-ResNet-101 · GTA -> Cityscapes | mIoU: 33.56 -> 42.53 | Improved |
| apply full method versus IBN-Net | ResNet-50 · GTA -> Cityscapes | mIoU: 29.64 -> 37.42 | Improved |
| use validation on target in the generalization method | Ours · GTA -> Cityscapes (VGG-16) | mIoU: 36.1 -> 38.6 | Improved |
| apply full method on SYNTHIA source | FCN-VGG-16 · SYNTHIA -> Cityscapes/BDDS/Mapillary | Gperf mIoU: 25.34 -> 32.41 | Improved |
Closest papers by meaning
- Constructing Self-Motivated Pyramid Curriculums for Cross-Domain Semantic Segmentation: A Non-Adversarial Approach (ICCV 2019 · similarity 0.88)
- Domain Stylization: A Fast Covariance Matching Framework Towards Domain Adaptation (TPAMI 2020 · similarity 0.88)
- Multi-source Domain Adaptation for Semantic Segmentation (NeurIPS 2019 · similarity 0.88)
- DIRL: Domain-Invariant Representation Learning for Generalizable Semantic Segmentation (AAAI 2022 · similarity 0.87)
- Towards Optimal Strategies for Training Self-Driving Perception Models in Simulation (NeurIPS 2021 · similarity 0.87)
- 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.
