Stagewise Unsupervised Domain Adaptation With Adversarial Self-Training for Road Segmentation of Remote-Sensing Images

Abstract

Semantic segmentation of remote sensing (RS) images is a challenging yet crucial task. While deep learning, particularly supervised learning with large-scale labeled datasets, has significantly advanced this field, acquiring high-quality labeled data is expensive and time-consuming. Additionally, variations in ground sampling distance, imaging equipment, and geographic differences cause domain shifts between datasets, which limit model performance across domains. Unsupervised domain adaptation (UDA) offers a solution by enabling models to learn from unlabeled target domain data while training on labeled source domain data. Recent self-supervised learning approaches using pseudo-label generation have shown promise in addressing domain discrepancies. Combining source and target images with their true and pseudo labels has proven effective in reducing domain bias. However, the use of pseudo-labeling for RS image segmentation is underexplored. Existing methods often rely on high-confidence pixel points as pseudo-labels, reducing supervision in low-confidence areas. Noise in pseudolabels further weakens the model's ability to learn target domain semantics. While some methods assign confidence weights, noisy pseudo-labels remain an issue. To address these limitations, we propose integrating contrastive learning into UDA, enhancing the model's capacity to capture semantic information by maximizing the similarity between augmented views of the same image. This provides additional supervision to improve performance in the target domain. Extensive experiments on key RS datasets (Potsdam, Vaihingen, LoveDA) demonstrate that our SimSeg method outperforms existing approaches, achieving state-of-theart results. Visualization and quantitative analyses confirm its superior ability to learn from the target domain. The code is available at https://github.com/woldier/SiamSeg.

Hyperspectral Classification via Global-Local Hierarchical Weighting Fusion Network

“…2) Size of Spatial Neighborhood Patch: We evaluate the impact of the spatial neighborhood patch size on the classification accuracy in detail by setting the spatial neighborhood patch λ size range of [17][18][19][20][21][22][23][24][25][26][27][28][29][30][31] with a step size of 2. As shown in Fig.…”

“…Deep learning (DL) captures the advanced features of the original data adaptively through a hierarchical structure. As a powerful feature extraction tool [18], it has been successfully applied in the field of remote sensing [19]- [22]. Chen et al [23] proposed a combination of stacked autoencoder and principal component analysis (PCA) to extract spectral features.…”

FMWDCT: Foreground Mixup Into Weighted Dual-Network Cross Training for Semisupervised Remote Sensing Road Extraction

“…In addition, some new encoder-decoder networks [32], [33] are also proposed for road extraction. RoadDA [34] was present to address the DS issue in road extraction task via adversarial self-training.…”

Cross-Domain Classification of Multisource Remote Sensing Data Using Fractional Fusion and Spatial-Spectral Domain Adaptation

“…Incorporating LiDAR reduces data-level per-pixel spectral shift caused by illumination variation, but there exist feature-level disparities between source and target scenes. With the aid of information from the source domain and the related features, the target domain can be adapted with the source domain and classified by domain adaptation (DA) techniques [27]- [29]. Inspired by the traditional DA method in machine learning, DA technology has been introduced to HSI classification to reduce the data shift and distribution bias [30]- [32].…”