Research map / Label-efficient and robust learning / Domain adaptation and generalization
SPIGAN: Privileged Adversarial Learning from Simulation
Kuan-Hui Lee, German Ros, Jie Li, Adrien Gaidon
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 simulator depth as privileged information (ablation) | SPIGAN on Cityscapes, 16 classes · Cityscapes val, 320×640 | mean IoU: 31.1 -> 34.7 | Improved |
| Add simulator depth as privileged information (ablation) | SPIGAN on Cityscapes, 16 classes · Cityscapes val, 512×1024 | mean IoU: 35.8 -> 36.8 | Improved |
| Add simulator depth as privileged information (ablation) | SPIGAN · Cityscapes, 7 classes | mIoU: 46.0 -> 53.4 | Improved |
| Add simulator depth as privileged information (ablation) | SPIGAN · Vistas, 7 classes | mIoU: 26.8 -> 44.1 | Improved |
| Add perceptual regularization (ablation) | SPIGAN-no-PI · Cityscapes, 7 classes | mIoU: 37.9 -> 46.0 | Improved |
| Add perceptual regularization (ablation) | SPIGAN-no-PI · Vistas, 7 classes | mIoU: 22.7 -> 26.8 | Improved |
| Add simulator depth as privileged information (ablation) | SPIGAN · Cityscapes, 7 classes | negative transfer rate: 0.16 -> 0.09 | Improved |
| Add simulator depth as privileged information (ablation) | SPIGAN · Vistas, 7 classes | negative transfer rate: 0.80 -> 0.42 | Improved |
| Adapt from synthetic images to the target domain with SPIGAN | SPIGAN · Cityscapes, 7 classes | mIoU: 36.3 -> 53.4 | Improved |
| Adapt from synthetic images to the target domain with SPIGAN | SPIGAN · Vistas, 7 classes | mIoU: 39.8 -> 44.1 | Improved |
Later papers that compared with this method
5 later accepted papers report SPIGAN as a baseline in their results tables; 5 of them report a better number on most of the results they share with it.
- Unsupervised Domain Adaptation for Semantic Segmentation using Depth Distribution (NeurIPS 2022 · better on 4 of 4 shared results)
- BiMaL: Bijective Maximum Likelihood Approach to Domain Adaptation in Semantic Scene Segmentation (ICCV 2021 · better on 3 of 3 shared results)
- Domain Adaptive Semantic Segmentation With Self-Supervised Depth Estimation (ICCV 2021 · better on 1 of 1 shared result)
- Geometric Unsupervised Domain Adaptation for Semantic Segmentation (ICCV 2021 · better on 1 of 1 shared result)
- DADA: Depth-Aware Domain Adaptation in Semantic Segmentation (ICCV 2019 · better on 5 of 5 shared results)
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
- Learning From Synthetic Data: Addressing Domain Shift for Semantic Segmentation (CVPR 2018 · similarity 0.90)
- Unsupervised Pixel-Level Domain Adaptation With Generative Adversarial Networks (CVPR 2017 · similarity 0.88)
- Conditional Generative Adversarial Network for Structured Domain Adaptation (CVPR 2018 · similarity 0.88)
- Domain transfer through deep activation matching (ECCV 2018 · similarity 0.87)
- Category-Level Adversarial Adaptation for Semantic Segmentation using Purified Features (TPAMI 2021 · similarity 0.86)
- An Adversarial Perturbation Oriented Domain Adaptation Approach for Semantic Segmentation (AAAI 2020 · 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.
