Research map / Label-efficient and robust learning / Domain adaptation and generalization
Penalizing Top Performers: Conservative Loss for Semantic Segmentation Adaptation
Xinge Zhu, Hui Zhou, Ceyuan Yang, Jianping Shi, Dahua Lin
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 the GAN output-space adversarial component (ablation) | FCN8s · GTAV -> Cityscapes | mIoU: 30.0 -> 34.4 | Improved |
| replace cross entropy with Conservative Loss (ablation) | FCN8s+GAN · GTAV -> Cityscapes | mIoU: 34.4 -> 38.1 | Improved |
| use warm start (cross entropy first) for Conservative Loss (ablation) | FCN8s+GAN+CL · GTAV -> Cityscapes | mIoU: 35.2 -> 38.1 | Improved |
| cold-start Conservative Loss instead of prior method (ablation) | FCN8s+GAN · GTAV -> Cityscapes | mIoU: 28.9 -> 35.2 | Improved |
| set logarithm base a to e instead of 2 (ablation) | Conservative Loss · GTAV -> Cityscapes | mIoU: 37.5 -> 38.1 | Improved |
| set logarithm base a to e instead of 4 (ablation) | Conservative Loss · GTAV -> Cityscapes | mIoU: 36.8 -> 38.1 | Improved |
| set lambda to 5 instead of 1 (ablation) | Conservative Loss · GTAV -> Cityscapes | mIoU: 37.2 -> 38.1 | Improved |
| set lambda to 5 instead of 20 (ablation) | Conservative Loss · GTAV -> Cityscapes | mIoU: 37.8 -> 38.1 | No clear effect |
| replace cross entropy with Focal Loss (ablation) | FCN8s+GAN · GTAV -> Cityscapes | mIoU: 34.4 -> 35.8 | Improved |
| replace Focal Loss with Conservative Loss (ablation) | FCN8s+GAN · GTAV -> Cityscapes | mIoU: 35.8 -> 38.1 | Improved |
| use piecewise cubic Loss3 instead of Conservative Loss (ablation) | FCN8s+GAN · GTAV -> Cityscapes | mIoU: 37.8 vs 38.1 | No clear effect |
| use homogeneous cubic Loss1 instead of cross entropy (ablation) | FCN8s+GAN · GTAV -> Cityscapes | mIoU: 34.4 -> 36.5 | Improved |
Closest papers by meaning
- ADVENT: Adversarial Entropy Minimization for Domain Adaptation in Semantic Segmentation (CVPR 2019 · similarity 0.89)
- On the Road to Online Adaptation for Semantic Image Segmentation (CVPR 2022 · similarity 0.89)
- Learning From Synthetic Data: Addressing Domain Shift for Semantic Segmentation (CVPR 2018 · similarity 0.89)
- SemiDAViL: Semi-supervised Domain Adaptation with Vision-Language Guidance for Semantic Segmentation (CVPR 2025 · similarity 0.89)
- Pulling Target to Source: A New Perspective on Domain Adaptive Semantic Segmentation (IJCV 2024 · similarity 0.89)
- Towards Unsupervised Domain Bridging via Image Degradation in Semantic Segmentation (NeurIPS 2025 · 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.
