Search

Boundary information is essential for the semantic segmentation of remote sensing images. However, most existing methods were designed to establish strong contextual information while losing detailed information, making it challenging to extract and recover boundaries accurately. In this article, a boundary-sensitive network (BSNet) is proposed to address this problem via dynamic hybrid gradient convolution (DHGC) and coordinate sensitive attention (CSA). Specifically, in the feature extraction stage, we propose DHGC to replace vanilla convolution (VC), which adaptively aggregates one VC kernel and two gradient convolution kernels (GCKs) into a new operator to enhance boundary information extraction. The GCKs are proposed to explicitly encode boundary information, which is inspired by traditional Sobel operators. In the feature recovery stage, the CSA is introduced. This module is used to reconstruct the sharp and detailed segmentation results by adaptively modeling the boundary information and long-range dependencies in the low-level features as the assistance of high-level features. Note that DHGC and CSA are plug-and-play modules. We evaluate the proposed BSNet on three public datasets: the ISPRS 2-D semantic labeling Vaihingen, the Potsdam benchmark, and the iSAID dataset. The experimental results indicate that BSNet is a highly effective architecture that produces sharper predictions around object boundaries and significantly improves the segmentation accuracy. Our method demonstrates superior performance on the Vaihingen, the Potsdam benchmark, and the iSAID dataset in terms of the mean $F_{1}$ , with improvements of 4.6%, 2.3%, and 2.4% over strong baselines, respectively. The code and models will be made publicly available. [ABSTRACT FROM AUTHOR]

The terrain segmentation in polarimetric synthetic aperture radar (PolSAR) images is an important task for image interpretation. Since the speckle noise and complex scattering mechanism exist in SAR images, the classification results achieved by traditional methods appear fragmented. Gradually, deep-learning-based methods are proposed to solve this problem. However, only the amplitude data in the SAR image is utilized, which limits the classification precision. In this letter, a novel method based on a dual-attention fusion network (DAFN) is presented. DAFN is mainly composed of a two-way structure encoder for feature extraction and the attention-based fusion module. Considering the terrain characteristic and the SAR imaging mechanism, the introduction of the polarization information in DAFN increases the discrimination of different categories, which contributes to the consistent and accurate fine-grained classification results. To demonstrate the effectiveness of the proposed method, the corresponding experiments are done based on a GaoFen-3 satellite full-polarization SAR data set, in which the superior performance in terrain segmentation is obtained. [ABSTRACT FROM AUTHOR]

Current unsupervised domain adaptation (UDA) techniques in semantic segmentation effectively decrease the domain discrepancy between the labeled source domain and unlabeled target domain, thereby enhancing the model's pixel-wise discriminative capability for target domain data. However, in remote sensing images (RSIs), our study uncovers that these approaches perform poorly in the presence of class distribution inconsistencies between the source and target domains. In this work, we propose a one-stage mean teacher framework with a novel auxiliary prototype classifier, named MTA, to address this issue. Specifically, the teacher model assigns pseudo labels at pixel level for target samples and captures knowledge from the student model via exponential moving average (EMA). With labeled source samples and target samples that have pseudo labels, the student model can alleviate the divergence between the source and target domains. In addition, the auxiliary prototype classifier (APC) reduces the performance degradation in the parametric softmax classifier of the student model caused by class distribution divergence. We also propose a prototype computation scheme to obtain each class prototype in the APC. Specifically, we build a memory bank for each class of the two domains to store feature embeddings dynamically. Then, we compute the class prototype by applying the clustering algorithm on memory banks corresponding to the class. Meanwhile, the APC reduces the intra-class domain discrepancy by optimizing the cross-entropy loss, which brings each class feature distribution of the two domains closer to the class prototype. The experimental results on RSIs UDA semantic segmentation tasks show the superiority of our approach over comparative methods. [ABSTRACT FROM AUTHOR]