Road Segmentation for Remote Sensing Images Using Adversarial Spatial Pyramid Networks

• Published in: IEEE transactions on geoscience and remote sensing 2021-06, Vol.59 (6), p.4673-4688
• Main Authors: Shamsolmoali, Pourya, Zareapoor, Masoumeh, Zhou, Huiyu, Wang, Ruili, Yang, Jie
• Format: Article
• Language: English
• Subjects: Accuracy; Adaptation models; Adversarial network; Annotations; Datasets; domain adaptation; Domains; Feature extraction; Feature maps; feature pyramid (FP); Generative adversarial networks; Generators; Image processing; Image quality; Image segmentation; Model accuracy; Optimization; Performance enhancement; Remote sensing; remote sensing (RS) images; road segmentation; Roads; Semantics; Task analysis; Training
• ISSN: 0196-2892; 1558-0644
• DOI: 10.1109/TGRS.2020.3016086

Road extraction in remote sensing images is of great importance for a wide range of applications. Because of the complex background, and high density, most of the existing methods fail to accurately extract a road network that appears correct and complete. Moreover, they suffer from either insufficient training data or high costs of manual annotation. To address these problems, we introduce a new model to apply structured domain adaption for synthetic image generation and road segmentation. We incorporate a feature pyramid (FP) network into generative adversarial networks to minimize the difference between the source and target domains. A generator is learned to produce quality synthetic images, and the discriminator attempts to distinguish them. We also propose a FP network that improves the performance of the proposed model by extracting effective features from all the layers of the network for describing different scales' objects. Indeed, a novel scale-wise architecture is introduced to learn from the multilevel feature maps and improve the semantics of the features. For optimization, the model is trained by a joint reconstruction loss function, which minimizes the difference between the fake images and the real ones. A wide range of experiments on three data sets prove the superior performance of the proposed approach in terms of accuracy and efficiency. In particular, our model achieves state-of-the-art 78.86 IOU on the Massachusetts data set with 14.89M parameters and 86.78B FLOPs, with 4\times fewer FLOPs but higher accuracy (+3.47% IOU) than the top performer among state-of-the-art approaches used in the evaluation.