Research map / Label-efficient and robust learning / Domain adaptation and generalization
Conditional Generative Adversarial Network for Structured Domain Adaptation
Read the paperCheck 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 |
|---|---|---|---|
| add conditional feature generation (ablation) | FCN-8s semantic segmentation adapted from SYNTHIA · Cityscapes val | mean IoU: 17.1 -> 41.2 | Helped |
| replace Skip Pooling with the conditional generator (ablation) | FCN-8s semantic segmentation adapted from SYNTHIA · Cityscapes val | mean IoU: 22.7 -> 41.2 | Helped |
| add conditional feature generation (ablation) | FCN-8s semantic segmentation adapted from GTA · Cityscapes val | mean IoU: 20.5 -> 44.5 | Helped |
| replace Skip Pooling with the conditional generator (ablation) | FCN-8s semantic segmentation adapted from GTA · Cityscapes val | mean IoU: 24.9 -> 44.5 | Helped |
| add a noise channel to the conditional generator (ablation) | FCN-8s semantic segmentation adapted from SYNTHIA · Cityscapes val | mean IoU: 40.7 -> 41.2 | Helped |
| add a noise channel to the conditional generator (ablation) | FCN-8s semantic segmentation adapted from GTA · Cityscapes val | mean IoU: 43.2 -> 44.5 | Helped |
| use generator input features from progressively higher convolutional layers instead of Conv1 (ablation) | conditional generator for structured domain adaptation · Cityscapes val | mean IoU | Hurt |
| increase the number of residual blocks in the conditional generator (ablation) | structured domain adaptation from SYNTHIA and GTA · Cityscapes val | mean IoU | Helped |
| add more synthetic training images (ablation) | structured domain adaptation from SYNTHIA and GTA · Cityscapes val | IoU | Helped |
| adapt from GTA rather than SYNTHIA | the proposed structured domain adaptation method · Cityscapes val | mean IoU: 41.2 -> 44.5 | Helped |
| use the proposed conditional GAN adaptation rather than NoAdapt | FCN-8s, adapting from SYNTHIA · Cityscapes val | mean IoU: 17.4 -> 41.2 | Helped |
| use the proposed conditional GAN adaptation rather than NoAdapt | FCN-8s, adapting from GTA · Cityscapes val | mean IoU: 21.1 -> 44.5 | Helped |
Closest papers by meaning
- Learning From Synthetic Data: Addressing Domain Shift for Semantic Segmentation (CVPR 2018 · similarity 0.93)
- Structured Generative Adversarial Networks (NeurIPS 2017 · similarity 0.90)
- Generate to Adapt: Aligning Domains Using Generative Adversarial Networks (CVPR 2018 · similarity 0.89)
- Unsupervised Pixel-Level Domain Adaptation With Generative Adversarial Networks (CVPR 2017 · similarity 0.89)
- Fully Convolutional Adaptation Networks for Semantic Segmentation (CVPR 2018 · similarity 0.88)
- Unlocking Pre-trained Image Backbones for Semantic Image Synthesis (CVPR 2024 · 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.