Building contour extraction from high-resolution remote sensing images is a basic task for the reasonable planning of regional construction. Recently, building segmentation methods based on the U-Net network have become popular as they largely improve the segmentation accuracy by applying ‘skip connection’ to combine high-level and low-level feature information more effectively. Meanwhile, researchers have demonstrated that introducing an attention mechanism into U-Net can enhance local feature expression and improve the performance of building extraction in remote sensing images. In this paper, we intend to explore the effectiveness of the primeval attention gate module and propose the novel Attention Gate Module (AG) based on adjusting the position of ‘Resampler’ in an attention gate to Sigmoid function for a building extraction task, and a novel Attention Gates U network (AGs-Unet) is further proposed based on AG, which can automatically learn different forms of building structures in high-resolution remote sensing images and realize efficient extraction of building contour. AGs-Unet integrates attention gates with a single U-Net network, in which a series of attention gate modules are added into the ‘skip connection’ for suppressing the irrelevant and noisy feature responses in the input image to highlight the dominant features of the buildings in the image. AGs-Unet improves the feature selection of the attention map to enhance the ability of feature learning, as well as paying attention to the feature information of small-scale buildings. We conducted the experiments on the WHU building dataset and the INRIA Aerial Image Labeling dataset, in which the proposed AGs-Unet model is compared with several classic models (such as FCN8s, SegNet, U-Net, and DANet) and two state-of-the-art models (such as PISANet, and ARC-Net). The extraction accuracy of each model is evaluated by using three evaluation indexes, namely, overall accuracy, precision, and intersection over union. Experimental results show that the proposed AGs-Unet model can improve the quality of building extraction from high-resolution remote sensing images effectively in terms of prediction performance and result accuracy.

Automatic building extraction based on high-resolution aerial imagery is an important challenge with a wide range of practical applications. One of the mainstream methods for extracting buildings from high-resolution images is deep learning because of its excellent deep feature extraction capability. However, existing models suffer from the problems of hollow interiors of some buildings and blurred boundaries. Furthermore, the increase in remote sensing image resolution has also led to rough segmentation results. To address these issues, we propose a generative adversarial segmentation network (ASGASN) for pixel-level extraction of buildings. The segmentation network of this framework adopts an asymmetric encoder–decoder structure. It captures and aggregates multiscale contextual information using the ASPP module and improves the classification and localization accuracy of the network using the global convolutional block. The discriminator network is an adversarial network that correctly discriminates the output of the generator and ground truth maps and computes multiscale $L_1$ loss by fusing multiscale feature mappings. The segmentation network and the discriminator network are trained alternately on the WHU building dataset and the China typical cities building dataset. Experimental results show that the proposed ASGASN can accurately identify different types of buildings and achieve pixel-level high accuracy extraction of buildings. Additionally, compared to available deep learning models, ASGASN also achieved the highest accuracy performance (89.4% and 83.6% IoU on these two datasets, respectively).

Deep learning technology, such as fully convolutional networks (FCNs), have shown competitive performance in the automatic extraction of buildings from high-resolution aerial images (HRAIs). However, there are problems of over-segmentation and internal cavity in traditional FCNs used for building extraction. To address these issues, this paper proposes a new building graph convolutional network (BGC-Net), which optimizes the segmentation results by introducing the graph convolutional network (GCN). The core of BGC-Net includes two major modules. One is an atrous attention pyramid (AAP) module, obtained by fusing the attention mechanism and atrous convolution, which improves the performance of the model in extracting multi-scale buildings through multi-scale feature fusion; the other is a dual graph convolutional (DGN) module, the build of which is based on GCN, which improves the segmentation accuracy of object edges by adding long-range contextual information. The performance of BGC-Net is tested on two high spatial resolution datasets (Wuhan University building dataset and a Chinese typical city building dataset) and compared with several state-of-the-art networks. Experimental results demonstrate that the proposed method outperforms several state-of-the-art approaches (FCN8s, DANet, SegNet, U-Net, ARC-Net, BAR-Net) in both visual interpretation and quantitative evaluations. The BGC-Net proposed in this paper has better results when extracting the completeness of buildings, including boundary segmentation accuracy, and shows great potential in high-precision remote sensing mapping applications.

The traditional deep learning networks in remote sensing image building segmentation have problems such as incomplete internal extraction, low accuracy of edge segmentation, and difficult prediction of irregular targets. Based on this, this paper proposes the network SCA-Net, which fuses the dense connection feature pyramid (DCFP) module and the spatial and channel attention mechanisms (SCA) module, for high precision automatic extraction of building contours. First, the DCFP module is used to obtain a large range of multi-scale contextual information without sacrificing the spatial resolution of features, thus improving the recognition rate of small-sized buildings. Second, SCA module is used to enhance feature learning ability of building pixels and suppress irrelevant feature information input. Finally, a fully convolutional asymmetric encoder-decoder structure is designed to obtain the final building segmentation results by gradually fusing the features and decoding them. The results on the WHU dataset as well as the CHN dataset show that the extraction of buildings using this method has good extraction results and high accuracy performance, outperforming several comparable advanced networks.