# Learning from Synthetic Data: Addressing Domain Shift for Semantic Segmentation

arXiv: https://arxiv.org/abs/1711.06969 · HTML: https://mlsanity.com/paper/1711.06969/

## Key facts

- Cited by 10 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 19 Nov 2017.

## Built on by (most cited first, 10 of 10)

- [Learning to Relate Depth and Semantics for Unsupervised Domain Adaptation](https://mlsanity.com/paper/2105.07830/) (2021)
- [Diffusion-based Image Translation with Label Guidance for Domain Adaptive Semantic Segmentation](https://mlsanity.com/paper/2308.12350/) (2023)
- [Bi-directional Contrastive Learning for Domain Adaptive Semantic Segmentation](https://mlsanity.com/paper/2207.10892/) (2022)
- [Exploiting Image Translations via Ensemble Self-Supervised Learning for Unsupervised Domain Adaptation](https://mlsanity.com/paper/2107.06235/) (2021)
- [Semi-supervised dry herbage mass estimation using automatic data and synthetic images](https://mlsanity.com/paper/2110.13719/) (2021)
- [Sparse-based Domain Adaptation Network for OCTA Image Super-Resolution Reconstruction](https://mlsanity.com/paper/2207.11882/) (2022)
- [A Batch Normalization Classifier for Domain Adaptation](https://mlsanity.com/paper/2103.11642/) (2021)
- [Foreground Object Structure Transfer for Unsupervised Domain Adaptation](https://mlsanity.com/paper/2109.06543/) (2021)
- [FSID: Fully Synthetic Image Denoising via Procedural Scene Generation](https://mlsanity.com/paper/2212.03961/) (2022)
- [Exploring the Landscape for Generative Sequence Models for Specialized Data Synthesis](https://mlsanity.com/paper/2411.01929/) (2024)

## Abstract (arXiv)

Visual Domain Adaptation is a problem of immense importance in computer vision. Previous approaches showcase the inability of even deep neural networks to learn informative representations across domain shift. This problem is more severe for tasks where acquiring hand labeled data is extremely hard and tedious. In this work, we focus on adapting the representations learned by segmentation networks across synthetic and real domains. Contrary to previous approaches that use a simple adversarial objective or superpixel information to aid the process, we propose an approach based on Generative Adversarial Networks (GANs) that brings the embeddings closer in the learned feature space. To showcase the generality and scalability of our approach, we show that we can achieve state of the art results on two challenging scenarios of synthetic to real domain adaptation. Additional exploratory experiments show that our approach: (1) generalizes to unseen domains and (2) results in improved alignment of source and target distributions.