# Rethinking Unsupervised Domain Adaptation for Semantic Segmentation

arXiv: https://arxiv.org/abs/2207.00067 · HTML: https://mlsanity.com/paper/2207.00067/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 27 of its 36 references that are in the map.
- First posted on arXiv 30 Jun 2022.

## Builds on (most cited first, 27 of 27)

- [PyTorch: An Imperative Style, High-Performance Deep Learning Library](https://mlsanity.com/paper/1912.01703/) (2019)
- [The Cityscapes Dataset for Semantic Urban Scene Understanding](https://mlsanity.com/paper/1604.01685/) (2016)
- [DeepLab: Semantic Image Segmentation with Deep Convolutional Nets, Atrous Convolution, and Fully Connected CRFs](https://mlsanity.com/paper/1606.00915/) (2016)
- [Unsupervised Domain Adaptation by Backpropagation](https://mlsanity.com/paper/1409.7495/) (2014)
- [Playing for Data: Ground Truth from Computer Games](https://mlsanity.com/paper/1608.02192/) (2016)
- [ADVENT: Adversarial Entropy Minimization for Domain Adaptation in Semantic Segmentation](https://mlsanity.com/paper/1811.12833/) (2018)
- [Learning to Adapt Structured Output Space for Semantic Segmentation](https://mlsanity.com/paper/1802.10349/) (2018)
- [Rethinking Few-Shot Image Classification: a Good Embedding Is All You Need?](https://mlsanity.com/paper/2003.11539/) (2020)
- [Confidence Regularized Self-Training](https://mlsanity.com/paper/1908.09822/) (2019)
- [Bidirectional Learning for Domain Adaptation of Semantic Segmentation](https://mlsanity.com/paper/1904.10620/) (2019)
- [Diverse Image-to-Image Translation via Disentangled Representations](https://mlsanity.com/paper/1808.00948/) (2018)
- [Frustratingly Simple Few-Shot Object Detection](https://mlsanity.com/paper/2003.06957/) (2020)
- [Rectifying Pseudo Label Learning via Uncertainty Estimation for Domain Adaptive Semantic Segmentation](https://mlsanity.com/paper/2003.03773/) (2020)
- [Unsupervised Intra-domain Adaptation for Semantic Segmentation through Self-Supervision](https://mlsanity.com/paper/2004.07703/) (2020)
- [Progressive Feature Alignment for Unsupervised Domain Adaptation](https://mlsanity.com/paper/1811.08585/) (2018)
- [DLOW: Domain Flow for Adaptation and Generalization](https://mlsanity.com/paper/1812.05418/) (2018)
- [Instance Adaptive Self-Training for Unsupervised Domain Adaptation](https://mlsanity.com/paper/2008.12197/) (2020)
- [Deep Reconstruction-Classification Networks for Unsupervised Domain Adaptation](https://mlsanity.com/paper/1607.03516/) (2016)
- [Classes Matter: A Fine-grained Adversarial Approach to Cross-domain Semantic Segmentation](https://mlsanity.com/paper/2007.09222/) (2020)
- [Few-Shot Segmentation Without Meta-Learning: A Good Transductive Inference Is All You Need?](https://mlsanity.com/paper/2012.06166/) (2020)
- [Cross-domain Object Detection through Coarse-to-Fine Feature Adaptation](https://mlsanity.com/paper/2003.10275/) (2020)
- [Minimal-Entropy Correlation Alignment for Unsupervised Deep Domain Adaptation](https://mlsanity.com/paper/1711.10288/) (2017)
- [Tune it the Right Way: Unsupervised Validation of Domain Adaptation via Soft Neighborhood Density](https://mlsanity.com/paper/2108.10860/) (2021)
- [Alleviating Semantic-level Shift: A Semi-supervised Domain Adaptation Method for Semantic Segmentation](https://mlsanity.com/paper/2004.00794/) (2020)
- [Semi-supervised Domain Adaptation based on Dual-level Domain Mixing for Semantic Segmentation](https://mlsanity.com/paper/2103.04705/) (2021)
- [SceneAdapt: Scene-based domain adaptation for semantic segmentation using adversarial learning](https://mlsanity.com/paper/2006.10386/) (2020)
- [Cross-Region Domain Adaptation for Class-level Alignment](https://mlsanity.com/paper/2109.06422/) (2021)

## Abstract (arXiv)

Unsupervised domain adaptation (UDA) adapts a model trained on one domain (called source) to a novel domain (called target) using only unlabeled data. Due to its high annotation cost, researchers have developed many UDA methods for semantic segmentation, which assume no labeled sample is available in the target domain. We question the practicality of this assumption for two reasons. First, after training a model with a UDA method, we must somehow verify the model before deployment. Second, UDA methods have at least a few hyper-parameters that need to be determined. The surest solution to these is to evaluate the model using validation data, i.e., a certain amount of labeled target-domain samples. This question about the basic assumption of UDA leads us to rethink UDA from a data-centric point of view. Specifically, we assume we have access to a minimum level of labeled data. Then, we ask how much is necessary to find good hyper-parameters of existing UDA methods. We then consider what if we use the same data for supervised training of the same model, e.g., finetuning. We conducted experiments to answer these questions with popular scenarios, GTA5, SYNTHIArightarrowCityscapes. We found that i) choosing good hyper-parameters needs only a few labeled images for some UDA methods whereas a lot more for others; and ii) simple finetuning works surprisingly well; it outperforms many UDA methods if only several dozens of labeled images are available.