Building Edges Research Articles

Bias variance Toeplitz Matrix based Shift Invariance classifier for building detection from satellite images

Buildings are a predominant category in man made objects of satellite images. As far as issues like building planning and shape of the building in satellite images plays a paramount aspect. Also, with the contemporary and swift deployment of technology, images are becoming more precise. Moreover, when the image correctness, the distinction in the composition and its features is increasingly obvious hence building feature identification with the aid of artificial selection has become a complicated task. Also, as the building edges are frequently adjoining to other features, there results in misclassification between final building extraction results and original buildings. Hence, novelty in building detection is of considerable importance to enhance the building detection with precise results with minimum classification (i.e., false positive rate). In this work an ideal classification method called, Toeplitz Matrix Convolutional Neural Network-based Shift Invariance (TMCNN-SI) for precise building detection from satellite images is proposed. The TMCNN-SI method is divided into three layers, namely, input layer, three hidden layers and output layer. The input layer by employing the Bias Variance Decomposition-based Generalization function obtains the feature point selection for further processing. In the first hidden layer, a Convolutional Kernel function to the convolved features is applied to generate feature map. In the second hidden layer, Toeplitz Matrix-based Shift Invariance function is applied with the objective of reducing the dimensionality involved in the generated feature map. In the third hidden layer, a Log Fusion function is applied for binary classification with minimum loss. Finally, the resultant target class is obtained in the output layer. The outline of the proposed TMCNN-SI method is studied in terms of classification time, classification accuracy and false positive rate. The radical results show that the TMCNN-SI approach bestows efficiently when considered with respect to classification.

How building layout properties influence pedestrian route choice and route recall

How pedestrians find and choose routes in buildings is a fundamental research topic that is immediately relevant to building design and safety. However, few studies have explored the relationship between pedestrian route choice and building layout in a systematic way. Here, we introduce a method based on spatial network theory for generating buildings with various layout properties. We conduct a virtual experiment with over 200 participants and many generated buildings to investigate how layout properties influence different aspects of pedestrian route choice. Our findings suggest route recall is worse in buildings that have more connections and possible routes, even when the overall size of buildings and length of routes is kept constant. Pedestrians also prefer more regular building layouts and are more likely to adopt the heuristic of walking along the outer edges of buildings the less regular they are and the more connections they have.

A Multiscale and Multitask Deep Learning Framework for Automatic Building Extraction

Detecting buildings, segmenting building footprints, and extracting building edges from high-resolution remote sensing images are vital in applications such as urban planning, change detection, smart cities, and map-making and updating. The tasks of building detection, footprint segmentation, and edge extraction affect each other to a certain extent. However, most previous works have focused on one of these three tasks and have lacked a multitask learning framework that can simultaneously solve the tasks of building detection, footprint segmentation and edge extraction, making it difficult to obtain smooth and complete buildings. This study proposes a novel multiscale and multitask deep learning framework to consider the dependencies among building detection, footprint segmentation, and edge extraction while completing all three tasks. In addition, a multitask feature fusion module is introduced into the deep learning framework to increase the robustness of feature extraction. A multitask loss function is also introduced to balance the training losses among the various tasks to obtain the best training results. Finally, the proposed method is applied to open-source building datasets and large-scale high-resolution remote sensing images and compared with other advanced building extraction methods. To verify the effectiveness of multitask learning, the performance of multitask learning and single-task training is compared in ablation experiments. The experimental results show that the proposed method has certain advantages over other methods and that multitask learning can effectively improve single-task performance.

Planning and Design Method of Multiangle Ecological Building Edge Space under the Background of Rural Revitalization

Under the background of rural revitalization, in order to realize the planning and design of ecological building edge space, a multi-perspective ecological building edge space planning and design method based on remote sensing image edge segmentation is proposed. The remote sensing visual detection of ecological buildings is realized by fusing multiscale features and multisource scene remote sensing images, and the extracted remote sensing image feature points are calibrated to extract the location information, texture features, super-resolution edge information features, and different levels of change features of the spatial distribution of the edge of ecological buildings. The background difference detection model of an ecological building remote sensing image is established, and the distance of the centroid of the corresponding level is calculated through frame dynamic planning and differential image clustering. Combined with the edge contour detection method of ecological building remote sensing image, the edge space planning and design are realized. The simulation results show that this method has higher accuracy in planning and better accuracy in detecting the contour of ecological building edge space and improves the dynamic planning and positioning ability of multi-perspective ecological building edge space distribution.

CA-BASNet: A Building Extraction Network in High Spatial Resolution Remote Sensing Images

Aiming at the problems of holes, misclassification, and rough edge segmentation in building extraction results from high spatial remote sensing images, a coordinate attention mechanism fusion network based on the BASNet network (CA-BASNet) is designed for building extraction in high spatial remote sensing images. Firstly, the deeply supervised encoder–decoder network was used to create a rough extract of buildings; secondly, to make the network pay more attention to learning building edge features, the mixed loss function composed of binary cross entropy, structural similarity and intersection-over-union was introduced into the network training process; finally, the residual optimization module of fusion coordinate attention mechanism was used for post-processing to realize the fine extraction of buildings from high spatial resolution remote sensing images. Experiments on the WHU building dataset show that the proposed network can achieve mIoU of 93.43%, mPA of 95.86%, recall of 98.79%, precision of 90.13% and F1 of 91.35%. Compared with the existing semantic segmentation networks, such as PSPNet, SegNet, DeepLapV3, SE-UNet, and UNet++, the accuracy of the proposed network and the integrity of object edge segmentation are significantly improved, which proves the effectiveness of the proposed network.

Cropland encroachment detection via dual attention and multi-loss based building extraction in remote sensing images.

The United Nations predicts that by 2050, the world’s total population will increase to 9.15 billion, but the per capita cropland will drop to 0.151°hm2. The acceleration of urbanization often comes at the expense of the encroachment of cropland, the unplanned expansion of urban area has adversely affected cultivation. Therefore, the automatic extraction of buildings, which are the main carriers of urban population activities, in remote sensing images has become a more meaningful cropland observation task. To solve the shortcomings of traditional building extraction methods such as insufficient utilization of image information, relying on manual characterization, etc. A U-Net based deep learning building extraction model is proposed and named AttsegGAN. This study proposes an adversarial loss based on the Generative Adversarial Network in terms of training strategy, and the additionally trained learnable discriminator is used as a distance measurer for the two probability distributions of ground truth Pdata and prediction Pg. In addition, for the sharpness of the building edge, the Sobel edge loss based on the Sobel operator is weighted and jointly participated in the training. In WHU building dataset, this study applies the components and strategies step by step, and verifies their effectiveness. Furthermore, the addition of the attention module is also subjected to ablation experiments and the final framework is determined. Compared with the original, AttsegGAN improved by 0.0062, 0.0027, and 0.0055 on Acc, F1, and IoU respectively after adopting all improvements. In the comparative experiment. AttsegGAN is compared with state-of-the-arts including U-Net, DeeplabV3+, PSPNet, and DANet on both WHU and Massachusetts building dataset. In WHU dataset, AttsegGAN achieved 0.9875, 0.9435, and 0.8907 on Acc, F1, and IoU, surpassed U-Net by 0.0260, 0.1183, and 0.1883, respectively, demonstrated the effectiveness of the proposed components in a similar hourglass structure. In Massachusetts dataset, AttsegGAN also surpassed state-of-the-arts, achieved 0.9395, 0.8328, and 0.7130 on Acc, F1, and IoU, respectively, it improved IoU by 0.0412 over the second-ranked PSPNet, and it was 0.0025 and 0.0101 higher than the second place in Acc and F1.

Advanced-Technological UAVs-Based Enhanced Reconstruction of Edges for Building Models

Accurate building models are widely used in the construction industry in the digital era. UAV cameras combined with image-based reconstruction provide a low-cost technology for building modeling. Most existing reconstruction methods operate on point clouds, while massive points reduce computational efficiency, and the accumulated error of point position often distorts building edges. This paper introduces an innovative 3D reconstruction method, Edge3D, that recovers building edges in the form of 3D lines. It employs geometry constraints and progressive screening technology to improve the robustness and precision of line segment matching. An innovative bundle adjustment strategy based on endpoints is designed to reduce the global reprojection error. Edges were tested on challenging real-world image sets, and matching precisions of 96% and 94% were achieved on the two image sets, respectively, with good reconstruction results. The proposed approach reconstructs building edges using a small number of lines instead of massive points, which contributes to the rapid reconstruction of building contour construction and obtaining accurate models, serving as an important foundation for the promotion of construction advancement.

A Robust Automatic Method to Extract Building Facade Maps from 3D Point Cloud Data

Extracting facade maps from 3D point clouds is a fast and economical way to describe a building’s surface structure. Existing methods lack efficiency, robustness, and accuracy, and depend on many additional features such as point cloud reflectivity and color. This paper proposes a robust and automatic method to extract building facade maps. First, an improved 3D Hough transform is proposed by adding shift vote and 3D convolution of the accumulator to improve computational efficiency and reduce peak fuzziness and dependence on the step selection. These modifications make the extraction of potential planes fast and accurate. Second, the coplane and vertical plane constraints are introduced to eliminate pseudoplanes and nonbuilding facades. Then, we propose a strategy to refine the potential facade and to achieve the accurate calibration and division of the adjacent facade boundaries by clustering the refined point clouds of the facade. This process solves the problem where adjoining surfaces are merged into the same surface in the traditional method. Finally, the extracted facade point clouds are converted into feature images. Doors, windows, and building edges are accurately extracted via deep learning and digital image processing techniques, which combine to achieve accurate extraction of building facades. The proposed method was tested on the MLS and TLS point cloud datasets, which were collected from different cities with different building styles. Experimental results confirm that the proposed method decreases computational burden, improves efficiency, and achieves the accurate differentiation of adjacent facade boundaries with higher accuracy compared with the traditional method, verifying the robustness of the method. Additionally, the proposed method uses only point cloud geometry information, effectively reducing data requirements and acquisition costs.

CD-TransUNet: A Hybrid Transformer Network for the Change Detection of Urban Buildings Using L-Band SAR Images

The change detection of urban buildings is currently a hotspot in the research area of remote sensing, which plays a vital role in urban planning, disaster assessments and surface dynamic monitoring. SAR images have unique characteristics compared with traditional optical images, mainly including abundant image information and large data volume. However, the majority of currently used SAR images for the detection of changes in buildings have the problems of missing the detection of small buildings and poor edge segmentation. Therefore, this paper proposes a new approach based on deep learning for changing building detection, which we called CD-TransUNet. It should be noted that CD-TransUNet is an end-to-end encoding–decoding hybrid Transformer model that combines the UNet and Transformer. Additionally, to enhance the precision of feature extraction and to reduce the computational complexity, the CD-TransUNet integrates coordinate attention (CA), atrous spatial pyramid pooling (ASPP) and depthwise separable convolution (DSC). In addition, by sending the differential images to the input layer, the CD-TransUNet can focus more on building changes over a large scale while ignoring the changes in other land types. At last, we verify the effectiveness of the proposed method using a pair of ALOS-2(L-band) acquisitions, and the comparative experimental results obtained from other baseline models show that the precision of the CD-TransUNet is much higher and the Kappa value can reach 0.795. Furthermore, the low missed alarms and the accurate building edge reflect that the proposed method is more appropriate for building changing detection tasks.

The Impact of Street Trees on a Typical Urban Canyon in Eastern Cairo Region

The trend of upgrading urban areas in Egypt is expanding across the region for better urban mobility patterns. Meanwhile, the concerns towards climate change and creating resilient communities are evolving to avoid the dramatic consequences on urban environments. Street trees offer great benefit in climate moderation, particularly in hot arid regions as in Cairo. In this work, different spatial arrangements and locations for evergreen, deciduous trees and palms at NW street canyon are tested using ENVI-met numerical simulation tool. It is aimed to maximize the cooling effect of tree shade on street facades and attain better pedestrian thermal comfort in peak summer times. This ultimately gives better guidance for urban planners and policymakers towards an improved planting design pattern. Comparing simulation results, it is concluded that evergreen trees located at the building edge can provide up to 6°C cooler walls and up to 3.8°C when placed at the building centre. Whilst other tree species have lower cooling effects. Also, these trees have a great role in attaining better thermal comfort levels for pedestrians at the street level. It is recommended to wisely choose suitable tree species and arrangements to maximize the cooling benefits in the urban environment.

Suggestive Data Annotation for CNN-Based Building Footprint Mapping Based on Deep Active Learning and Landscape Metrics

Convolutional neural network (CNN)-based very high-resolution (VHR) image segmentation has become a common way of extracting building footprints. Despite publicly available building datasets and pre-trained CNN models, it is still necessary to prepare sufficient labeled image tiles to train CNN models from scratch or update the parameters of pre-trained CNN models to extract buildings accurately in real-world applications, especially the large-scale building extraction, due to differences in landscapes and data sources. Deep active learning is an effective technique for resolving this issue. This study proposes a framework integrating two state-of-the-art (SOTA) models, U-Net and DeeplabV3+, three commonly used active learning strategies, (i.e., margin sampling, entropy, and vote entropy), and landscape characterization to illustrate the performance of active learning in reducing the effort of data annotation, and then understand what kind of image tiles are more advantageous for CNN-based building extraction. The framework enables iteratively selecting the most informative image tiles from the unlabeled dataset for data annotation, training the CNN models, and analyzing the changes in model performance. It also helps us to understand the landscape features of iteratively selected image tiles via active learning by considering building as the focal class and computing the percent, the number of patches, edge density, and landscape shape index of buildings based on labeled tiles in each selection. The proposed method was evaluated on two benchmark building datasets, WHU satellite dataset II and WHU aerial dataset. Models in each iteration were trained from scratch on all labeled tiles. Experimental results based on the two datasets indicate that, for both U-Net and DeeplabV3+, the three active learning strategies can reduce the number of image tiles to be annotated and achieve good model performance with fewer labeled image tiles. Moreover, image tiles with more building patches, larger areas of buildings, longer edges of buildings, and more dispersed building distribution patterns were more effective for model training. The study not only provides a framework to reduce the data annotation efforts in CNN-based building extraction but also summarizes the preliminary suggestions for data annotation, which could facilitate and guide data annotators in real-world applications.

OPTIMIZATION OF DISTRIBUTION OF INSULATION ON THERMAL INSULATION SHELL OF ENERGY EFFICIENT BUILDINGS

Nowadays, the issue of improving the energy efficiency of new buildings and thermal modernization of existing housing is acute. This can be done due to the rational insulation of external enclosing structures, the optimal redistribution of insulation on the edges of buildings, taking into account the thermal energy impact of the environment. 
 Existing regulations regulate the minimum level of heat transfer resistance through the enclosing structures of buildings, but do not take into account such features of thermal energy impact of the environment as the level of heat from solar radiation and wind. Thus, the issue of optimal redistribution of the normative amount of insulation between the faces of the building's thermal insulation shell remains legitimate, taking into account the thermal impact of the environment in order to minimize heat loss through enclosing structures, which will increase energy efficiency of buildings. 
 To solve the above problem, the method of optimizing the heat transfer resistance of translucent and opaque enclosing structures under different conditions and the structure of optimizing the redistribution of insulation between the faces of enclosing structures is proposed. 

Codification study of wind-induced loads on canopies attached to mid-rise buildings

Wind-induced loads on canopies attached to low-rise buildings have been investigated in the past and formed the basis for design recommendations in the latest wind standards and building codes of practice. The scope of this study was to extend the investigation to canopies that are attached to taller buildings and generate additional recommendations on the design load calculation of such non-structural attachments.Various geometrical parameters such as canopy length, tilt angle, installation location and building height along with wind direction were considered in the experimental protocol. Therefore, several configurations of a 1:20 scale model were tested in an atmospheric boundary layer wind tunnel. Pressures on both the upper and lower surfaces of the canopy were measured for different wind directions and surface and net mean and peak pressure coefficients are presented. The results indicated that an important flow mechanism influencing the peak pressures on the attached canopies is the turbulence generated by the canopy and building edges at the wind directions parallel to the canopy's parent wall. From a codification point of view, the area-averaged coefficients versus effective area curves were obtained and enveloped, considering all model configurations, to generate design area-averaged coefficients curves. These codified design curves can be incorporated into wind standards and building codes of practice for the purpose of designing attached canopies to mid-rise buildings.

IMPROVING PAIRWISE DSM WITH 3SGM: A SEMANTIC SEGMENTATION FOR SGM USING AN AUTOMATICALLY REFINED NEURAL NETWORK

Abstract. The amount of very high resolution optical satellite images at our disposal is continuously increasing. Besides, associated satellite programs often come with high revisit rates and geometric properties that allow for either opportunistic or by-design 3D stereo reconstruction. Digital Surface Models (DSM) computed from these satellite images offer new possibilities. In the past, the high revisit rate has largely benefited glacier monitoring studies. Now, DSM with increased resolution provided on urban areas can be used for smart city applications as well. However, most of these require 3D modeling of buildings with level of details ranging from 0 to 2. This is where the need for better reconstructed buildings inside DSM arises. Indeed, building edges and corners tend to be smoothed and softened by the stereo matching step of a DSM computation pipeline. This undesired behavior can mostly be linked to the difficult task of optimizing the Disparity Space Image, thus finding good balance between smoothing untextured areas while conserving sharp discontinuities where needed. In this paper, we show how the optimization can benefit from an input building semantic segmentation. We also provide a method to create it from a very high satellite image in epipolar geometry using a convolutional neural network. To help our network generalize well on unseen areas we propose an interactive learning method based on clicked annotations. Eventually, we show that annotations can be automatically created, hence removing the need for an operator and making our solution suitable for operational conditions.

Active Contour Building Segmentation Model based on Convolution Neural Network

In high-resolution remote sensing images, artificial features on the surface account for a large proportion. In artificial features, buildings, as special artificial features, buildings have different shapes. They are easily affected by light, so it takes a long time to extract using traditional image segmentation methods. It can't effectively design feature engineering to depict the high-dimensional features of the target building. We propose an active contour model based on a convolution neural network, which integrates the prior knowledge and constraints of active contour model, such as continuity of boundary, smooth edge, and geometric characteristics of buildings, into the learning process of convolution neural network to realize the close unity of ACM and CNN. According to our work, a fundamental end-to-end trainable image segmentation framework which is composed of convolution neural network (CNN) and ACM with learnable parameters is implemented, the problem of semantic segmentation of buildings in aerial images was dealt with, the model was evaluated on the publicly available dataset called Vaihingen, and some parameters were explained. In building semantics, the active contour model based on a convolution neural network has good performance.

World Settlement Footprint 3D - A first three-dimensional survey of the global building stock

Settlements, and in particular cities, are at the center of key future challenges related to global change and sustainable development. Widely used indicators to assess the efficiency and sustainability of settlement development are the compactness and density of the built-up area. However, at global scale, a temporally consistent and spatially detailed survey of the distribution and concentration of the building stock – meaning the total area and volume of buildings within a defined spatial unit or settlement, commonly referred to as building density – does not yet exist. To fill this data and knowledge gap, an approach was developed to map key characteristics of the world's building stock in a so far unprecedented level of spatial detail for every single settlement on our planet. The resulting World Settlement Footprint 3D dataset quantifies the fraction, total area, average height, and total volume of buildings for a measuring grid with 90 m cell size. The World Settlement Footprint 3D is generated using a modified version of the World Settlement Footprint human settlements mask derived from Sentinel-1 and Sentinel-2 satellite imagery at 10 m spatial resolution, in combination with 12 m digital elevation data and radar imagery collected by the TanDEM-X mission. The underlying, automated processing framework includes three basic workflows: one estimating the mean building height based on an analysis of height differences along potential building edges, a second module determining the building fraction and total building area within each 90 m cell, and a third part combining the height information and building area in order to determine the average height and total built-up volume at 90 m gridding. Optionally, a simple 3D building model (level of detail 1) can be generated for regions where data on the building footprints is available. A comprehensive validation campaign based on 3D building models obtained for 19 regions (~86,000 km2) and street-view samples indicating the number of floors for >130,000 individual buildings in 15 additional cities documents that the novel World Settlement Footprint 3D data provides valuable and, for the first time, globally consistent information on key characteristics of the building stock in both, large urban agglomerations as well as small-scale rural settlements. Thus, the new dataset represents a promising baseline dataset for a wide range of previously impossible environmental, socioeconomic, and climatological studies worldwide.

Quasi-Dense Matching Algorithm for Close-Range Image Combined With Feature Line Constraint

Point-based sparse or dense matching can typically obtain satisfactory 3D point clouds of general contour features, but the deformation problem at the edges of artificial objects is prominent. Thus, a quasi-dense matching algorithm for close-range images combined with feature line constraint is proposed in this study. The method utilizes reliable matched points to construct the initial Delaunay triangulation and then optimizes the triangulation using the matched feature line. On this basis, iterative quasi-dense matching based on triangulation constraint is implemented. The process achieves matching with the center of the inscribed circle of each triangle as the seed point and growing matching. Two sets of stereo image pairs acquired using smartphones and four sets of sequence images provided by public datasets are selected for quasi-dense matching experiments. The comparison of results of constraint matching of the two triangulations before and after optimization as well as the matching results obtained via VisualSFM software demonstrated that the 3D point cloud obtained via quasi-dense matching with feature line constraint presents better results at the edges of buildings, thereby confirming the effectiveness of the proposed algorithm.

Boundary Enhancement Semantic Segmentation for Building Extraction From Remote Sensed Image

Image processing via convolutional neural network (CNN) has been developed rapidly for remote sensing technology. Moreover, techniques for accurately extracting building footprints from remote sensed images have attracted considerable interest owing to their wide variety of common applications, including monitoring natural disasters and urban development. Extraction of building footprints can be performed easily by semantic segmentation using U-Net-like CNN architectures. However, obtaining precise boundaries of segmentation masks remains challenging due to various impediments surrounding target objects. In this study, we propose a method to elaborate edges of buildings detected in remote sensed images to enhance the boundaries of segmentation masks. The proposed method adopts <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">holistically nested edge detection (HED)</i> , which extracts edge features at an encoder of a given architecture. In the proposed <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">boundary enhancement (BE) module</i> , an extracted edge and segmentation mask are combined, sharing mutual information. To enable the proposed method efficiently to adapt to a wide variety of conditions, we design a distinctive approach adopting a HED unit and BE module, which is applicable to various semantic segmentation networks containing encoder-decoder structures. Experiments were conducted on five different datasets (DeepGlobe, Urban3D, WHU [high-resolution (HR), low-resolution (LR)], and Massachusetts). The results demonstrate that our proposed approaches improved on the performance of prior methods for extracting building footprints. Comparative experiments were conducted on various backbone architectures including U-Net, ResUNet++, TernausNet, and U-shape spatial pyramid pooling (USPP) to ensure the effectiveness of the proposed method. Based on various evaluation metrics and qualitative analysis, our results show that the proposed method achieved improved performance compared with prior methods for all datasets and backbone networks.

Structure-Aware Weakly Supervised Network for Building Extraction From Remote Sensing Images

The use of fully supervised deep learning methods to extract buildings from remote sensing images has shown excellent performance, which requires large amounts of training data with laborious per-pixel labeling. Compared with pixelated intensive labeling, it is much easier to label data using scribbles, which only takes few seconds for one image. In this paper, a one-stage structure-aware weakly supervised network (SAWSN) for building extraction is proposed, and it learns from easily accessible scribbles rather than from densely annotated ground truth. Firstly, to solve the problem that direct training with scribble labels will lead to poor building structures, an auxiliary edge detection task is introduced to localize building edges explicitly. Secondly, a structure aware scribble extension module (SASEM) is designed to recover building structures from scribbles through effective utilization of edge features. Finally, an edge-structure-aware loss is proposed to limit the scope of the restored structure. We perform extensive experiments on three newly labeled benchmark building extraction datasets (WHU, ISPRS Potsdam, and Vaihingen). Experimental results show that our method achieved 91.72%, 92.83%, and 92.22% of F1 using the ISPRS Vaihingen, Potsdam, and WHU datasets, respectively, and outperformed the state-of-the-art scribble-based weakly supervised methods by 3.27% of <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">IoU</i> .

Uncertainty, Edge, and Reverse-Attention Guided Generative Adversarial Network for Automatic Building Detection in Remotely Sensed Images

Despite recent advances in deep-learning-based semantic segmentation, automatic building detection from remotely sensed imagery is still a challenging problem owing to large variability in the appearance of buildings across the globe. The errors occur mostly around the boundaries of the building footprints, in shadow areas, and when detecting buildings whose exterior surfaces have reflectivity properties that are very similar to those of the surrounding regions. To overcome these problems, we propose a generative adversarial network-based segmentation framework with uncertainty attention unit and refinement module embedded in the generator. The refinement module, composed of edge and reverse-attention units, is designed to refine the predicted building map. The edge attention enhances the boundary features to estimate building boundaries with greater precision, and the reverse attention allows the network to explore the features missing in the previously estimated regions. The uncertainty attention unit assists the network in resolving uncertainties in classification. As a measure of the power of our approach, as of December 4, 2021, it ranks at the second place on DeepGlobe’s public leaderboard despite the fact that main focus of our approach—refinement of the building edges—does not align exactly with the metrics used for leaderboard rankings. Our overall <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"><tex-math notation="LaTeX">$F$</tex-math></inline-formula> 1-score on DeepGlobe’s challenging dataset is 0.745. We also report improvements on the previous-best results for the challenging INRIA validation dataset for which our network achieves an overall IoU of 81.28% and an overall accuracy of 97.03%. Along the same lines, for the official INRIA test dataset, our network scores 77.86% and 96.41% in overall IoU and accuracy. We have also improved upon the previous best results on two other datasets: the WHU building dataset and the Massachusetts buildings dataset.