BSNet: Dynamic Hybrid Gradient Convolution Based Boundary-Sensitive Network for Remote Sensing Image Segmentation

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.

I. Introduction

With the rapid development of remote sensing technology, an increasing number of remote sensing images with high resolution (e.g., 5–10 cm) can be obtained. In these images, small objects, such as cars and trees, can be clearly observed, which makes pixel-level semantic segmentation possible. Image semantic segmentation allocates pixelwise semantic labels in an image, segmenting the image into several regions based on semantic units. Semantic segmentation of remote sensing images has a wide range of applications, including environmental monitoring [1], postdisaster reconstruction [2], agriculture [3], forestry [4], and urban planning [5], [6]. Traditional image semantic segmentation methods [7]–[10] manually design feature classifiers but do not achieve satisfactory performance due to their limited feature extraction and data fitting abilities. These shortcomings have prompted researchers to search for more stable and efficient methods.