Awesome Segmentation Domain Adaptation

Awesome Segmentation Domain Adaptation

If you notice any result or the public code that has not been included in this page, please connect Zhedong Zheng without hesitation to add the method. You are welcomed! or create pull request.

Priorities are given to papers whose codes are published.

Benchmark results

Results on the two standard synthetic-to-real benchmarks for this task. All numbers are mIoU on the Cityscapes validation set, each method at its best reported configuration, taken from the comparison tables of the original papers. Where a method is quoted by several papers the value may differ by about 1 mIoU depending on the evaluation protocol, see the notes below.

GTA5 → Cityscapes (19 classes)

SYNTHIA → Cityscapes (16 classes)

Notes

• “†” marks methods that additionally train with depth data. Their scores are not directly comparable with the RGB-only methods.
• The two tables are not comparable with each other. GTA5 → Cityscapes is evaluated over 19 classes while SYNTHIA → Cityscapes is evaluated over the 16 shared classes. For SYNTHIA the mIoU must be averaged over 16 classes; averaging over 19 under-reports the score by about 4 points.
• Both tables are a timeline rather than a strict ranking, because the protocol changes over time (backbone, crop size, single-scale or sliding-window inference). The backbone is the dominant factor: HRDA reaches 63.0 with DeepLabV2 and 73.8 with MiT-B5.
• Values come from the original papers or from the comparison tables of MIC (CVPR2023), DiDA (NeurIPS2025) and BLDA (ICML2025). The same method can differ by about 1 mIoU between reports (MIC is 75.9 in its own paper and 74.61 in CoPT), so a gap below about 1.5 mIoU should not be read as progress.
• GTA5 → Cityscapes is close to saturated: the best RGB-only result (77.1) is within 0.5 mIoU of supervised training on Cityscapes (77.6). Since 2024, most papers use GTA5 as the labelled source domain but target ACDC, Dark Zurich, BDD100K or Mapillary instead.

Papers

Arxiv

• Cross-Region Domain Adaptation for Class-level Alignment [14 Sep 2021]
• Contrastive Learning and Self-Training for Unsupervised Domain Adaptation in Semantic Segmentation [5 May 2021]
• ACDC: The Adverse Conditions Dataset with Correspondences for Semantic Driving Scene Understanding [29 April 2021]
• Domain Adaptive Semantic Segmentation with Self-Supervised Depth Estimation [28 Apr 2021]
• Class-Conditional Domain Adaptation on Semantic Segmentation [27 Nov 2019]
• Domain Bridge for Unpaired Image-to-Image Translation and Unsupervised Domain Adaptation [23 Oct 2019]
• Restyling Data: Application to Unsupervised Domain Adaptation [24 Sep 2019]
• Adversarial Learning and Self-Teaching Techniques for Domain Adaptation in Semantic Segmentation [2 Sep 2019]
• FCNs in the Wild: Pixel-level Adversarial and Constraint-based Adaptation [8 Dec 2016]

Journal

• Adaptive Boosting for Domain Adaptation: Towards Robust Predictions in Scene Segmentation [TIP 2022][code]
• Rectifying Pseudo Label Learning via Uncertainty Estimation for Domain Adaptive Semantic Segmentation [IJCV 2021]
• Affinity Space Adaptation for Semantic Segmentation Across Domains [TIP2020]
• Semantic-aware short path adversarial training for cross-domain semantic segmentation [Neurocomputing 2019]
• Weakly Supervised Adversarial Domain Adaptation for Semantic Segmentation in Urban Scenes [TIP 2019]

Conference

• RecycleLoRA: Rank-Revealing QR-Based Dual-LoRA Subspace Adaptation for Domain Generalized Semantic Segmentation [CVPR2026] [Code]
• Open-Vocabulary Domain Generalization in Urban-Scene Segmentation [CVPR2026] [Code]
• Towards Unsupervised Domain Bridging via Image Degradation in Semantic Segmentation [NeurIPS2025] [Code]
• Balanced Learning for Domain Adaptive Semantic Segmentation [ICML2025] [Code]
• Unlocking Constraints: Source-Free Occlusion-Aware Seamless Segmentation [ICCV2025] [Code]
• Exploiting Domain Properties in Language-Driven Domain Generalization for Semantic Segmentation [ICCV2025] [Code]
• FreeDNA: Endowing Domain Adaptation of Diffusion-Based Dense Prediction with Training-Free Domain Noise Alignment [ICCV2025] [Code]
• Stronger, Steadier & Superior: Geometric Consistency in Depth VFM Forges Domain Generalized Semantic Segmentation [ICCV2025] [Code]
• SemiDAViL: Semi-supervised Domain Adaptation with Vision-Language Guidance for Semantic Segmentation [CVPR2025] [Code]
• Cross-Domain Semantic Segmentation on Inconsistent Taxonomy using VLMs [ECCV2024] [Code]
• CoPT: Unsupervised Domain Adaptive Segmentation using Domain-Agnostic Text Embeddings [ECCV2024] [Code]
• HALO: Hyperbolic Active Learning for Semantic Segmentation under Domain Shift [ICML2024] [Code]
• Open-Set Domain Adaptation for Semantic Segmentation [CVPR2024] [Code]
• Style Blind Domain Generalized Semantic Segmentation via Covariance Alignment and Semantic Consistence Contrastive Learning [CVPR2024] [Code]
• Stable Neighbor Denoising for Source-free Domain Adaptive Segmentation [CVPR2024] [Code]
• Stronger, Fewer, & Superior: Harnessing Vision Foundation Models for Domain Generalized Semantic Segmentation [CVPR2024] [Code]
• Collaborating Foundation Models for Domain Generalized Semantic Segmentation [CVPR2024] [Code]
• A Simple Recipe for Language-guided Domain Generalized Segmentation [CVPR2024] [Code]
• ATTA: Anomaly-aware Test-Time Adaptation for Out-of-Distribution Detection in Segmentation [NeurIPS2023] [Code]
• EDAPS: Enhanced Domain-Adaptive Panoptic Segmentation [ICCV2023] [Code]
• CMDA: Cross-Modality Domain Adaptation for Nighttime Semantic Segmentation [ICCV2023] [Code]
• Texture Learning Domain Randomization for Domain Generalized Segmentation [ICCV2023] [Code]
• DiGA: Distil to Generalize and then Adapt for Domain Adaptive Semantic Segmentation [CVPR2023] [Code]
• HGFormer: Hierarchical Grouping Transformer for Domain Generalized Semantic Segmentation [CVPR2023] [Code]
• MIC: Masked Image Consistency for Context-Enhanced Domain Adaptation [CVPR2023] [Code]
• Transferring to Real-World Layouts: A Depth-aware Framework for Scene Adaptation (DCF) [ACM MM2024] [Code]
• PiPa: Pixel- and Patch-wise Self-supervised Learning for Domain Adaptative Semantic Segmentation [ACM MM2023] [Code]
• DAformer: Improving Network Architectures and Training Strategies for Domain-Adaptive Semantic Segmentation [CVPR2022] [Code]
• Class-balanced pixel-level self-labeling for domain adaptive semantic segmentation [CVPR2022] [Code]
• Undoing the damage of label shift for cross-domain semantic segmentation [CVPR2022] [Code]
• Adas: A direct adaptation strategy for multi-target domain adaptive semantic segmentation [[CVPR2022]]
• HRDA: Context-Aware High-Resolution Domain-Adaptive Semantic Segmentation [ECCV2022] [Code]
• DecoupleNet: Decoupled Network for Domain Adaptive Semantic Segmentation [ECCV2022] [Code]
• Prototypical Contrast Adaptation for Domain Adaptive Semantic Segmentation [ECCV2022] [Code]
• OnDA: Online Domain Adaptation for Semantic Segmentation in Ever-Changing Conditions [ECCV2022] [Code]
• D2ADA: Dynamic Density-aware Active Domain Adaptation for Semantic Segmentation [ECCV2022] [Code]
• TACS: Taxonomy Adaptive Cross-Domain Semantic Segmentation [ECCV2022] [Code]
• Style-Hallucinated Dual Consistency Learning for Domain Generalized Semantic Segmentation [ECCV2022] [Code]
• Semantic-Aware Domain Generalized Segmentation [CVPR2022] [Code]
• WildNet: Learning Domain Generalized Semantic Segmentation from the Wild [CVPR2022] [Code]
• Towards Fewer Annotations: Active Learning via Region Impurity and Prediction Uncertainty for Domain Adaptive Semantic Segmentation [CVPR2022] [Code]
• Generalize Then Adapt: Source-Free Domain Adaptive Semantic Segmentation [ICCV2021]
• Prototypical Pseudo Label Denoising and Target Structure Learning for Domain Adaptive Semantic Segmentation [CVPR2021]
• Improving Domain Generalization in Urban-Scene Segmentationvia Instance Selective Whitening [CVPR2021]
• Rethinking Ensemble-Distillation for Semantic Segmentation Based Unsupervised Domain Adaptation [CVPRW 2021]
• Instance Adaptive Self-Training for Unsupervised Domain Adaptation [ECCV 2020]
• Two-phase Pseudo Label Densification for Self-training based Domain Adaptation [ECCV 2020]
• Unsupervised Scene Adaptation with Memory Regularization in vivo [IJCAI 2020] [code]
• Adversarial Style Mining for One-Shot Unsupervised Domain Adaptation [NeurIPS 2020][Pytorch]
• Content-Consistent Matching for Domain Adaptive Semantic Segmentation [ECCV 2020] [Code]
• Differential Treatment for Stuff and Things: A Simple Unsupervised Domain Adaptation Method for Semantic Segmentation [CVPR 2020] [Code]
• Contextual-Relation Consistent Domain Adaptation for Semantic Segmentation[ECCV 2020]
• An Adversarial Perturbation Oriented Domain Adaptation Approach for Semantic Segmentation [AAAI2020]
• Category Anchor-Guided Unsupervised Domain Adaptation for Semantic Segmentation [NeurIPS2019]) [code]
• MLSL: Multi-Level Self-Supervised Learning for Domain Adaptation with Spatially Independent and Semantically Consistent Labeling [WACV2020]
• Guided Curriculum Model Adaptation and Uncertainty-Aware Evaluation for Semantic Nighttime Image Segmentation [ICCV2019]
• Constructing Self-motivated Pyramid Curriculums for Cross-Domain Semantic Segmentation: A Non-Adversarial Approach [ICCV2019] [Pytorch]
• SSF-DAN: Separated Semantic Feature Based Domain Adaptation Network for Semantic Segmentation [ICCV2019]
• Significance-aware Information Bottleneck for Domain Adaptive Semantic Segmentation [ICCV2019]
• Domain Adaptation for Semantic Segmentation with Maximum Squares Loss [ICCV2019] [Pytorch]
• Self-Ensembling with GAN-based Data Augmentation for Domain Adaptation in Semantic Segmentation [ICCV2019]
• DADA: Depth-aware Domain Adaptation in Semantic Segmentation [ICCV2019] [code]
• Domain Adaptation for Structured Output via Discriminative Patch Representations [ICCV2019 Oral] [Project]
• Not All Areas Are Equal: Transfer Learning for Semantic Segmentation via Hierarchical Region Selection [CVPR2019(Oral)(PDF Coming Soon)]
• CrDoCo: Pixel-level Domain Transfer with Cross-Domain Consistency [CVPR2019] [Project]
• Bidirectional Learning for Domain Adaptation of Semantic Segmentation [CVPR2019] [Pytorch]
• Learning Semantic Segmentation from Synthetic Data: A Geometrically Guided Input-Output Adaptation Approach [CVPR2019]
• All about Structure: Adapting Structural Information across Domains for Boosting Semantic Segmentation [CVPR2019] [Pytorch]
• DLOW: Domain Flow for Adaptation and Generalization [CVPR2019 Oral]
• Taking A Closer Look at Domain Shift: Category-level Adversaries for Semantics Consistent Domain Adaptation [CVPR2019 Oral] [Pytorch]
• ADVENT: Adversarial Entropy Minimization for Domain Adaptation in Semantic Segmentation [CVPR2019 Oral] [Pytorch]
• SPIGAN: Privileged Adversarial Learning from Simulation [ICLR2019]
• Penalizing Top Performers: Conservative Loss for Semantic Segmentation Adaptation [ECCV2018]
• Domain transfer through deep activation matching [ECCV2018]
• Unsupervised Domain Adaptation for Semantic Segmentation via Class-Balanced Self-Training [ECCV2018]
• DCAN: Dual channel-wise alignment networks for unsupervised scene adaptation [ECCV2018]
• Fully convolutional adaptation networks for semantic segmentation [CVPR2018]
• Learning to Adapt Structured Output Space for Semantic Segmentation [CVPR2018] [Pytorch]
• Conditional Generative Adversarial Network for Structured Domain Adaptation [CVPR2018]
• Learning From Synthetic Data: Addressing Domain Shift for Semantic Segmentation [CVPR2018]
• Curriculum Domain Adaptation for Semantic Segmentation of Urban Scenes [ICCV2017] [Journal Version]
• No more discrimina- tion: Cross city adaptation of road scene segmenters [ICCV2017]

Reference

• https://github.com/zhaoxin94/awesome-domain-adaptation#semantic-segmentation