# Exploiting Semantics in Adversarial Training for Image-Level Domain Adaptation

arXiv: https://arxiv.org/abs/1810.05852 · HTML: https://mlsanity.com/paper/1810.05852/

## Key facts

- Cited by 9 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 13 Oct 2018.

## Built on by (most cited first, 9 of 9)

- [Survey on Unsupervised Domain Adaptation for Semantic Segmentation for Visual Perception in Automated Driving](https://mlsanity.com/paper/2304.11928/) (2023)
- [Unsupervised Domain Adaptation via Domain-Adaptive Diffusion](https://mlsanity.com/paper/2308.13893/) (2023)
- [Exploiting the Complementarity of 2D and 3D Networks to Address Domain-Shift in 3D Semantic Segmentation](https://mlsanity.com/paper/2304.02991/) (2023)
- [Plugging Self-Supervised Monocular Depth into Unsupervised Domain Adaptation for Semantic Segmentation](https://mlsanity.com/paper/2110.06685/) (2021)
- [Shallow Features Guide Unsupervised Domain Adaptation for Semantic Segmentation at Class Boundaries](https://mlsanity.com/paper/2110.02833/) (2021)
- [Physics-informed Guided Disentanglement in Generative Networks](https://mlsanity.com/paper/2107.14229/) (2021)
- [ManiFest: Manifold Deformation for Few-shot Image Translation](https://mlsanity.com/paper/2111.13681/) (2021)
- [Learning Good Features to Transfer Across Tasks and Domains](https://mlsanity.com/paper/2301.11310/) (2023)
- [Leveraging Local Domains for Image-to-Image Translation](https://mlsanity.com/paper/2109.04468/) (2021)

## Abstract (arXiv)

Performance achievable by modern deep learning approaches are directly related to the amount of data used at training time. Unfortunately, the annotation process is notoriously tedious and expensive, especially for pixel-wise tasks like semantic segmentation. Recent works have proposed to rely on synthetically generated imagery to ease the training set creation. However, models trained on these kind of data usually under-perform on real images due to the well known issue of domain shift. We address this problem by learning a domain-to-domain image translation GAN to shrink the gap between real and synthetic images. Peculiarly to our method, we introduce semantic constraints into the generation process to both avoid artifacts and guide the synthesis. To prove the effectiveness of our proposal, we show how a semantic segmentation CNN trained on images from the synthetic GTA dataset adapted by our method can improve performance by more than 16% mIoU with respect to the same model trained on synthetic images.