MSAFNet: Multiscale Successive Attention Fusion Network for Water Body Extraction of Remote Sensing Images

Abstract

Water body extraction is a typical task in the semantic segmentation of remote sensing images (RSIs). Deep convolutional neural networks (DCNNs) outperform traditional methods in mining visual features; however, due to the inherent convolutional mechanism of the network, spatial details and abstract semantic representations at different levels are difficult to capture accurately at the same time, and then the extraction results decline to become suboptimal, especially on narrow areas and boundaries. To address the above-mentioned problem, a multiscale successive attention fusion network, named MSAFNet, is proposed to efficiently aggregate the multiscale features from two aspects. A successive attention fusion module (SAFM) is first devised to extract multiscale and fine-grained features of water bodies, while a joint attention module (JAM) is proposed to further mine salient semantic information by jointly modeling contextual dependencies. Furthermore, the multi-level features extracted by the above-mentioned modules are aggregated by a feature fusion module (FFM) so that the edges of water bodies are well mapped, directly improving the segmentation of various water bodies. Extensive experiments were conducted on the Qinghai-Tibet Plateau Lake (QTPL) and the Land-cOVEr Domain Adaptive semantic segmentation (LoveDA) datasets. Numerically, MSAFNet reached the highest accuracy on both QTPL and LoveDA datasets, including Kappa, MIoU, FWIoU, F1, and OA, outperforming several mainstream methods. Regarding the QTPL dataset, MSAFNet peaked at 99.14% and 98.97% in terms of F1 and OA. Although the LoveDA dataset is more challenging, MSAFNet retained the best performance, with F1 and OA being 97.69% and 95.87%. Additionally, visual inspections exhibited consistency with numerical evaluations.

A Spectral–Spatial Context-Boosted Network for Semantic Segmentation of Remote Sensing Images

“…For comparative analysis, eleven advanced approaches were selected: U-Net [20], DeepLabV3+ [22], DANet [30], ResUNet-a [48], DASSN [35], HCANet [36], RAANet [56], SCAttNet [55], A2-FPN [54], LANet [38], and SAPNet [57]. Notably, U-Net [20], DeepLabV3+ [22], and DANet [30] were initially developed for natural image segmentation, while ResUNeta [48], DASSN [35], HCANet [36], RAANet [56], SCAttNet [55], A2-FPN [54], LANet [38], MSAFNet [67], CLCFormer [68], and SAPNet [57] represent recent state-of-the-art methodologies specifically designed for RSI segmentation.…”

Editorial for Special Issue: “Monitoring Terrestrial Water Resource Using Multiple Satellite Sensors”

“…Wu et al [27] proposed a remote-sensing-based method for mapping channel activity using the MNDWI index combined with the Otsu method based on the Google Earth Engine (GEE) platform, applying the method with the Lower Yellow River as a case study. Lyu et al [28] developed a MSAFNet (multiscale successive attention fusion network) model for extracting water bodies from remote sensing images; they tested their method on the Qinghai-Tibet Plateau Lake (QTPL) and the Land-cOVEr Domain Adaptive semantic segmentation (LoveDA) datasets and achieved a high mapping accuracy (i.e., overall accuracy of 98.97% on the QTPL dataset and 95.87% on the LoveDA dataset). Lu et al [29] proposed a model for extracting river networks by combining DEM, a Landsat-derived global surface water occurrence (GSWO) dataset, and Sentinel-2 imagery, and they applied the model to conduct a case study across the Danjiangkou Reservoir Area, finding results consistent with the actual river network.…”

An Enhanced Dual-Stream Network Using Multi-Source Remote Sensing Imagery for Water Body Segmentation

“…This is accomplished through parallel channel-specific attention and spatial-specific attention, which enhance the differentiation between the foreground and background. MSAFNet [26] is a network that employs multiple attentional modules to extract both low-level and high-level local features. These multi-level features are then aggregated through the feature fusion module (FFM) to improve the mapping of water body edges, directly enhancing the segmentation of water bodies.…”