Building Polygons Research Articles

Second-order texton feature extraction and pattern recognition of building polygon cluster using CNN network

The cluster patterns of features in map space represent a comprehensive reflection of individual feature geometric attributes and their spatial adjacency relationships. These patterns also embody spatial cognition results under the Gestalt principle. Describing non-linear spatial cluster patterns as effective regular structures is one of the fundamental tasks in deep learning for recognizing feature cluster patterns. In this study, based on the concept of texture co-occurrence matrices from regular gray-scale images, we utilized Voronoi diagrams to construct the tessellation structure of building polygons. Built upon the foundation of first-order texton co-occurrence matrices, we established three-dimensional texton co-occurrence matrices for building polygons, considered five attributes of building size, shape, orientation, and density, and encompassed 64 different combinations of second-order neighboring directions. This matrix concretizes the latent Gestalt spatial characteristics of building polygon clusters into a three-dimensional sparse matrix. It is then used as an input vector to construct a deep convolutional neural network for recognizing building polygon cluster patterns. Through adjustments and optimizations of neural network structure and strategies, along with validation through practical case studies and comparisons with other models, we have demonstrated the effectiveness of the second-order texton co-occurrence matrix in describing the characteristics of building polygon clusters.

Exploring an unknown: Representative sample survey on structure and energy-related quality of the non-residential building stock in Germany

The non-residential building stock in Germany, as in many other countries, is not fully represented in any official statistics, in contrast to residential buildings. This is surprising given the economic significance of this asset and its importance as a source of greenhouse gases. The knowledge gap was closed in a representative sample survey providing statistically valid data on structural parameters and energy-related characteristics of the non-residential building stock in Germany. This became possible because by December 2014 authoritative building polygons from all German cadastres were available in a national database for the first time. These geospatial data, adjusted for topological inconsistencies and supplemented with owner information in an on-site screening, were used as a sampling frame in the previously unknown population of the non-residential buildings, a new approach in building stock statistics worldwide. While the geometry of all buildings can be derived from the geospatial data, energy-related characteristics and renovation activities have to be obtained by interviews with the owners of the relevant non-residential buildings in the sample. With this methodology, regular monitoring of building stocks at reasonable costs becomes possible. It also can be transferred to other states with similar geospatial data infrastructure, in particular to other EU Member States.

From lines to Polygons: Polygonal building contour extraction from High-Resolution remote sensing imagery

Automated extraction of polygonal building contours from high-resolution remote sensing images is important for various applications. However, it remains a difficult task to achieve automated extraction of polygonal buildings at the level of human delineation due to diverse building structures and imperfect image conditions. In this paper, we propose Line2Poly, an end-to-end approach that uses feature lines as geometric primitives to achieve polygonal building extraction by recovering topological relationships among these lines within an individual building. To extract building feature lines with precision, we adopt a two-stage strategy that combines Convolutional Neural Network (CNN) and transformer architectures. A CNN-based module extracts preliminary feature lines, which serve as positional priors for initializing positional queries in the subsequent transformer-based module. For polygonal building contour reconstruction, we devise a learnable polygon topology reconstruction module that predicts adjacency relationships among discrete lines, and integrates lines into building polygons. The resultant building polygons, based on feature lines, exhibit inherent regularity that aligns with manual labeling standards. Extensive experiments on the Vectorizing World Buildings dataset, the WHU aerial building dataset and the WHU-Mix (vector) dataset validate Line2Poly’s impressive performance in building feature line extraction and instance-level building detection. Moreover, Line2Poly’s predictions exhibit the highest level of concurrence with manual delineations, with over 83% agreement on the WHU aerial building test set and 68.7/59.7% on the WHU-Mix (vector) test set I and II, respectively.

Building Polygon Extraction from High-Resolution Remote Sensing Imagery Using Knowledge Distillation

Building polygons plays an important role in urban management. Although leveraging deep learning techniques for building polygon extraction offers advantages, the models heavily rely on a large number of training samples to achieve good generalization performance. In scenarios with small training samples, the models struggle to effectively represent diverse building structures and handle the complexity introduced by the background. A common approach to enhance feature representation is fine-tuning a pre-trained model on a large dataset specific to the task. However, the fine-tuning process tends to overfit the model to the task area samples, leading to the loss of generalization knowledge from the large dataset. To address this challenge and enable the model to inherit the generalization knowledge from the large dataset while learning the characteristics of the task area samples, this paper proposes a knowledge distillation-based framework called Building Polygon Distillation Network (BPDNet). The teacher network of BPDNet is trained on a large building polygon dataset containing diverse building samples. The student network was trained on a small number of available samples from the target area to learn the characteristics of the task area samples. The teacher network provides guidance during the training of the student network, enabling it to learn under the supervision of generalization knowledge. Moreover, to improve the extraction of buildings against the backdrop of a complex urban context, characterized by fuzziness, irregularity, and connectivity issues, BPDNet employs the Dice Loss, which focuses attention on building boundaries. The experimental results demonstrated that BPDNet effectively addresses the problem of limited generalization by integrating the generalization knowledge from the large dataset with the characteristics of the task area samples. It accurately identifies building polygons with diverse structures and alleviates boundary fuzziness and connectivity issues.

Building footprint extraction from aerial photogrammetric point cloud data using its geometric features

Unmanned aerial vehicles (UAVs) can quickly acquire high-resolution datasets. Generally, UAVs or drones have high-resolution optical cameras onboard to obtain aerial images. These images are processed to provide various output products, including point cloud, digital surface model (DSM), digital terrain model (DTM), and ortho-mosaiced image. This study uses point cloud data obtained from UAV data processing to extract buildings automatically. It utilizes the geometric features obtainable from the point cloud data in a defined neighbourhood to classify the point cloud data. Normalized DSM (nDSM) is also used as an input to identify above-ground features more accurately. Random Forest (RF) algorithm has been used to classify the point cloud data into different classes available in the dataset. After classification, buildings are separated from the point cloud data, and K-Means clustering is performed to segregate different building clusters. These clusters are rasterized, and morphological operations are applied to refine the building edges. Then the boundaries of the building clusters are identified to provide the vector data. Accuracy assessment of the automatically extracted shapes is done by comparing their area, perimeter, and centroid location to the reference building polygons generated through the total station survey. The methodology is tested over the dataset acquired through UAV. An open-source GUI (graphical user interface) based tool has been developed in Python to extract the vectorized building shapes from photogrammetric point cloud data and compute areas automatically. It will reduce manual interventions significantly and benefit many users, professionals and researchers.

Time-Use Data Modelling of Domestic, Commercial and Industrial Electricity Demand for the Scottish Islands

Achieving emissions reduction targets requires improved energy efficiency to avoid an oversized and excessively expensive electricity network. This can be analysed using hourly demand modelling that captures behaviour profiles, technology types, weather factors and building typologies. Numerous domestic models exist, but whole systems energy modelling, including commercial and industrial demand, are limited by data availability. Time-use survey data has typically been used to model domestic demand- in this work is expanded to also model commercial and industrial electricity-heating for the Scottish islands at an hourly and individual building level. This method is widely applicable for modelling whole system energy demand wherever time-use survey data are available. Combinatorial optimisation has been applied to generate a synthetic population, match individuals to properties and apply construction types to building polygons. SimStock is used for heating and lighting modelling. Validation of the model with 2016 data shows that it reflects longer term trends, with a monthly mean absolute percentage error (MAPE) of 1.6% and an R2 of 0.99. At the hourly level, the MAPE of 6.2% and R2 of 0.87 show the model captures variability needed to combine it with a supply-side model. Dataset accuracy, variability in the date recorded, missing data and unknown data correlations are discussed as causes for error. The model can be adapted for other regions and used to analyse the costs and benefits of energy efficiency measures with a supply-side generation model.

BCE-Net: Reliable building footprints change extraction based on historical map and up-to-date images using contrastive learning

Automatic and periodic recompiling of building databases with up-to-date high-resolution images has become a critical requirement for rapidly developing urban environments. However, the architecture of most existing approaches for change extraction attempts to learn features related to changes but ignores objectives related to buildings. This inevitably leads to the generation of significant pseudo-changes, due to factors such as seasonal changes in images and the inclination of building façades. To alleviate the above-mentioned problems, we developed a contrastive learning approach by validating historical building footprints against single up-to-date remotely sensed images. This contrastive learning strategy allowed us to inject the semantics of buildings into a pipeline for the detection of changes, which is achieved by increasing the distinguishability of features of buildings from those of non-buildings. In addition, to reduce the effects of inconsistencies between historical building polygons and buildings in up-to-date images, we employed a deformable convolutional neural network to learn offsets intuitively. In summary, we formulated a multi-branch building extraction method that identifies newly constructed and removed buildings, respectively. To validate our method, we conducted comparative experiments using the public Wuhan University building change detection dataset and a more practical dataset named SI-BU that we established. Our method achieved F1 scores of 93.99% and 70.74% on the above datasets, respectively. Moreover, when the data of the public dataset were divided in the same manner as in previous related studies, our method achieved an F1 score of 94.63%, which surpasses that of the state-of-the-art method. Code and datasets are available at https://vrlab.org.cn/~hanhu/projects/bcenet.

Joint semantic–geometric learning for polygonal building segmentation from high-resolution remote sensing images

As a fundamental task for geographical information updating, 3D city modeling, and other critical applications, the automatic extraction of building footprints from high-resolution remote sensing images has been substantially explored and received increasing attention over recent years. Among different types of building extraction methods, the polygonal segmentation methods produce vector building polygons that are in a more realistic format compared with those obtained from pixel-wise semantic labeling and contour-based methods. However, existing polygonal building segmentation methods usually require a perfect segmentation map and a complex post-processing procedure to guarantee the polygonization quality, or produce inaccurate vertex prediction results that suffer from wrong vertex sequence, self-intersections, fixed vertex quantity, etc. In our previous work, we have proposed a method for polygonal building segmentation from remote sensing images that addresses the above limitations of existing methods. In this paper, we propose PolyCity, which further extends and improves our previous work in terms of the application scenario, methodology design, and experimental results. Our proposed PolyCity contains the following three components: (1) a pixel-wise multi-task network for learning the semantic and geometric information via three tasks, i.e., building segmentation, boundary prediction, and edge orientation prediction; (2) a simple but effective vertex selection module (VSM), which effectively bridges the gap between pixel-wise and graph-based models via transforming the segmentation map into valid polygon vertices; (3) a graph-based vertex refinement network (VRN) for automatically adjusting the coordinates of VSM-generated valid polygon vertices, producing the final building polygons with more precise vertices. Results on three large-scale building extraction datasets demonstrate that our proposed PolyCity generates vector building footprints with more accurate vertices, edges, shapes, etc., achieving significant vertex score improvements while maintaining high segmentation and boundary scores compared with the current state-of-the-art. The code of PolyCity will be released at https://github.com/liweijia/polycity.

Distinguishing Buildings from Vegetation in an Urban-Chaparral Mosaic Landscape with LiDAR-Informed Discriminant Analysis

Identification of buildings from remotely sensed imagery in urban and suburban areas is a challenging task. Light detection and Ranging (LiDAR) provides an opportunity to accurately identify buildings by identification of planar surfaces. Dense vegetation can limit the number of light particles that reach the ground, potentially creating false planar surfaces within a vegetation stand. We present an application of discriminant analysis (a commonly used statistical tool in decision theory) to classify polygons (derived from LiDAR) as either buildings or a non-building planar surfaces. We conducted our analysis in southern Texas where thornscrub vegetation often prevents a LiDAR beam from fully penetrating the vegetation canopy in and around residential areas. Using discriminant analysis, we grouped potential building polygons into building and non-building classes using the point densities of ground, unclassified, and building points. Our technique was 95% accurate at distinguishing buildings from non-buildings. Therefore, we recommend its use in any locale where distinguishing buildings from surrounding vegetation may be affected by the proximity of dense vegetation to buildings.

High-Precision Single Building Model Reconstruction Based on the Registration between OSM and DSM from Satellite Stereos

For large-scale 3D building reconstruction, there have been several approaches to utilizing multi-view satellite imagery to produce a digital surface model (DSM) for height information and extracting building footprints for contour information. However, limited by satellite resolutions and viewing angles, the corresponding DSM and building footprints are sometimes of a low accuracy, thus generating low-accuracy building models. Though some recent studies have added GIS data to refine the contour of the building footprints, the registration errors between the GIS data and satellite images are not considered. Since OpenStreetMap (OSM) provides a high level of precision and complete building polygons in most cities worldwide, this paper proposes an automatic single building reconstruction method that utilizes a DSM from high-resolution satellite stereos, as well as building footprints from OSM. The core algorithm accurately registers the building polygons from OSM with the rasterized height information from the DSM. To achieve this goal, this paper proposes a two-step “coarse-to-fine registration” algorithm, with both steps being formulated into the optimization of energy functions. The coarse registration is optimized by separately moving the OSM polygons at fixed steps with the constraints of a boundary gradient, an interior elevation mean, and variance. Given the initial solution of the coarse registration, the fine registration is optimized by a genetic algorithm to compute the accurate translations and rotations between the DSM and OSM. Experiments performed in the Beijing/Shanghai region show that the proposed method can significantly improve the IoU (intersection over union) of the registration results by 69.8%/26.2%, the precision by 41.0%/15.5%, the recall by 41.0%/16.0%, and the F1-score by 42.7%/15.8%. For the registration, the method can reduce the translation errors by 4.656 m/2.815 m, as well as the rotation errors by 0.538°/0.228°, which indicates its great potential in smart 3D applications.

BuildMapper: A fully learnable framework for vectorized building contour extraction

Deep learning based methods have significantly boosted the study of automatic building extraction from remote sensing images. However, delineating vectorized and regular building contours like a human does remains very challenging, due to the difficulty of the methodology, the diversity of building structures, and the imperfect imaging conditions. In this paper, we propose the first end-to-end learnable building contour extraction framework, named BuildMapper, which can directly and efficiently delineate building polygons just as a human does. BuildMapper consists of two main components: 1) a contour initialization module that generates initial building contours; and 2) a contour evolution module that performs both contour vertex deformation and reduction, which removes the need for complex empirical post-processing used in existing methods. In both components, we provide new ideas, including a learnable contour initialization method to replace the empirical methods, dynamic predicted and ground truth vertex pairing for the static vertex correspondence problem, and a lightweight encoder for vertex information extraction and aggregation, which benefit a general contour-based method; and a well-designed vertex classification head for building corner vertices detection, which casts light on direct structured building contour extraction. We also built a suitable large-scale building dataset, the WHU-Mix (vector) building dataset, to benefit the study of contour-based building extraction methods. The extensive experiments conducted on the WHU-Mix (vector) dataset, the WHU dataset, and the CrowdAI dataset verified that BuildMapper can achieve a state-of-the-art performance, with a higher mask average precision (AP) and boundary AP than both segmentation-based and contour-based methods. We also confirmed that more than 60.0/50.8% of the building polygons predicted by BuildMapper in the WHU-Mix (vector) test sets I/II, 84.2% in the WHU building test set, and 68.3% in the CrowdAI test set are on par with the manual delineation level.

Estimation of the Photovoltaic Potential on Rooftops in the City of San Pedro Sula, Cortés, Honduras

Honduras is a country that is making a transition away from fossil fuels and towards sustainable energy sources. Currently in San Pedro Sula there is not an accurate data that represent an estimate of photovoltaic potential per district. Calculating an estimate of the photovoltaic potential is important to know how much decentralized energy can be replaced on the energy matrix by photovoltaic energy and thus be able to achieve an energy matrix that has 100% renewable energy in the country. In this study, we calculated an estimate of the rooftop solar power potential over ten out of the twenty districts in the city of San Pedro Sula using globally available solar radiation data from Meteonorm combined with a building polygon, the analysis of the photovoltaic potential was completed using two photovoltaic modules. The annual rooftop solar power potential in San Pedro Sula under Scenario A was of 2.04 GWp and for Scenario B was of 1.88 GWp. This approach of combining Meteonorm data with building polygons can be easily applied in other parts of the country. These findings can provide useful information for policymakers and contribute to local planning for cleaner renewable energy.

Multi-Task Edge Detection for Building Vectorization From Aerial Images

The extraction of building outline vectors is an essential task in supporting various applications. Although the recent development of deep learning based techniques has made advancements in the automation of this task, the accuracy and precision are insufficient due to errors caused by abundant noise and obstruction around buildings in aerial images. To better address this issue, this paper presents a new approach called the multi-task edge detection (MTED) for building vectorization with the following characteristics. First, instead of detecting building corner points that are very sensitive to noise effects, a deep learning based rotated bounding box detector is introduced for building edge detection to increase robustness to interference. Second, a multi-task learning strategy is designed to integrate building segmentation inside the METD framework to closely guide the edge detection using spatial context. Third, a simple yet effective geometry-guided post-processing method is designed to reconstruct vectorized building outlines based on the detected edges and learned building shape prior knowledge. The comparative experiments conducted on benchmark very-high-resolution optical aerial images indicate that the proposed approach can significantly outperform the state-of-the-arts in terms of vertex-based building outline accuracy metrics. With a test time of 58ms per building, this method enables efficient building polygon labeling in interactive mapping applications for building surveying and mapping. Code is available at https: //github.com/yifanthomaswu/MTED_framework.

Analyzing Satellite-Derived 3D Building Inventories and Quantifying Urban Growth towards Active Faults: A Case Study of Bishkek, Kyrgyzstan

Earth observation (EO) data can provide large scale, high-resolution, and transferable methodologies to quantify the sprawl and vertical development of cities and are required to inform disaster risk reduction strategies for current and future populations. We synthesize the evolution of Bishkek, Kyrgyzstan, which experiences high seismic hazard, and derive new datasets relevant for seismic risk modeling. First, the urban sprawl of Bishkek (1979–2021) was quantified using built-up area land cover classifications. Second, a change detection methodology was applied to a declassified KeyHole Hexagon (KH-9) and Sentinel-2 satellite image to detect areas of redevelopment within Bishkek. Finally, vertical development was quantified using multi-temporal high-resolution stereo and tri-stereo satellite imagery, which were used in a deep learning workflow to extract buildings footprints and assign building heights. Our results revealed urban growth of 139 km2 (92%) and redevelopment of ~26% (59 km2) of the city (1979–2021). The trends of urban growth were not reflected in all the open access global settlement footprint products that were evaluated. Building polygons that were extracted using a deep learning workflow applied to high-resolution tri-stereo (Pleiades) satellite imagery were most accurate (F1 score = 0.70) compared to stereo (WorldView-2) imagery (F1 score = 0.61). Similarly, building heights extracted using a Pleiades-derived digital elevation model were most comparable to independent measurements obtained using ICESat-2 altimetry data and field-measurements (normalized absolute median deviation < 1 m). Across different areas of the city, our analysis suggested rates of building growth in the region of 2000–10,700 buildings per year, which when combined with a trend of urban growth towards active faults highlights the importance of up-to-date building stock exposure data in areas of seismic hazard. Deep learning methodologies applied to high-resolution imagery are a valuable monitoring tool for building stock, especially where country-level or open-source datasets are lacking or incomplete.

A Progressive Simplification Method for Buildings Based on Structural Subdivision

Building simplification is an important research area in automatic map generalization. Up to now, many approaches have been proposed by scholars. However, in the continuous transformation of scales for buildings, keeping the main shape characteristics, area, and orthogonality of buildings are always the key and difficult points. Therefore, this paper proposes a method of progressive simplification for buildings based on structural subdivision. In this paper, iterative simplification is adopted, which transforms the problem of building simplification into the simplification of the minimum details of building outlines. Firstly, a top priority structure (TPS) is determined, which represents the smallest detail in the outline of the building. Then, according to the orthogonality and concave–convex characteristics, the TPS are classified as 62 subdivisions, which cover the local structure of the building polygon. Then, the subdivisions are divided into four simplification types. The building is simplified to eliminate the TPS continuously, retaining the right-angle characteristics and area as much as possible, until the results satisfy the constraints and rules of simplification. A topographic dataset (1:1 K) collected from Kadaster was used for our experiments. In order to evaluate the algorithm, many tests were undertaken, including tests of multi-scale simplification and simplification of typical buildings, which indicate that this method can realize multi-scale presentation of buildings. Compared with the existing simplification methods, the comparison results show that the proposed method can simplify buildings effectively, which has certain advantages in keeping shape characteristics, area, and rectangularity.

Estimation of Rooftop Solar Power Potential by Comparing Solar Radiation Data and Remote Sensing Data—A Case Study in Aichi, Japan

There have been significant advances in the shift from fossil-based energy systems to renewable energies in recent years. Decentralized solar photovoltaic (PV) is one of the most promising energy sources because of the availability of rooftop areas, ease of installation, and reduced cost of PV panels. The current modeling method using remote sensing data based on a geographic information system (GIS) is objective and accurate, but the analysis processes are complicated and time-consuming. In this study, we developed a method to estimate the rooftop solar power potential over a wide area using globally available solar radiation data from Solargis combined with a building polygon. Our study also utilized light detection and ranging (LiDAR) data and AW3D to estimate rooftop solar power potential in western Aichi, Japan, and the solar radiation was calculated using GIS. The estimation using LiDAR data took into account the slope and azimuth of rooftops. A regression analysis of the estimated solar power potential for each roof between the three methods was conducted, and the conversion factor 0.837 was obtained to improve the accuracy of the results from the Solargis data. The annual rooftop solar power potential of 3,351,960 buildings in Aichi Prefecture under Scenario A, B, and C was 6.92 × 107, 3.58 × 107, and 1.27 × 107 MWh/year, estimated using Solargis data after the adjustment. The estimated solar power potential under Scenario A could satisfy the total residential power demand in Aichi, revealing the crucial role of rooftop solar power in alleviating the energy crisis. This approach of combining Solargis data with building polygons can be easily applied in other parts of the world. These findings can provide useful information for policymakers and contribute to local planning for cleaner energy.

Automated 3D Reconstruction of LoD2 and LoD1 Models for All 10 Million Buildings of the Netherlands

In this paper, we present our workflow to automatically reconstruct three-dimensional (3D) building models based on two-dimensional building polygons and a lidar point cloud. The workflow generates models at different levels of detail (LoDs) to support data requirements of different applications from one consistent source. Specific attention has been paid to make the workflow robust to quickly run a new iteration in case of improvements in an algorithm or in case new input data become available. The quality of the reconstructed data highly depends on the quality of the input data and is monitored in several steps of the process. A 3D viewer has been developed to view and download the openly available 3D data at different LoDs in different formats. The workflow has been applied to all 10 million buildings of the Netherlands. The 3D service will be updated after new input data becomes available.

Building Classification Algorithm Based on Improved Hough Transform and Geographic Information System

Although determining the type of building can be useful for 3D reconstruction, image retrieval, and other applications, it is often difficult to do so automatically. A building classification algorithm using a geographic information system (GIS) is proposed in this paper. The typical adjustment of building polygon, local convex-concave simplification of large-area polygon contour, and local exaggeration of competing polygon groups merging in small-area polygon is the main topics of this paper. The preliminary outline area is obtained by analyzing the corner characteristics of two basic types of buildings in urban buildings using a large-scale morphological screen. The classification statistics of all kinds of corners are carried out using the improved Hough transform and the proposed line segment and corner optimization algorithm, and the automatic classification of flat-roofed and nonflat-roofed buildings photographed by digital cameras is realized. The experimental results show that the algorithm simplifies the polygons in a reasonable way while maintaining the block shape. Furthermore, the optimization algorithm proposed in this paper effectively eliminates the influence of false contours, allowing for high-accuracy building type judgment.

Analyzing the Behaviors of OpenStreetMap Volunteers in Mapping Building Polygons Using a Machine Learning Approach

Mapping as an action in volunteered geographic information is complex in light of the human diversity within the volunteer community. There is no integrated solution that models and fixes all data heterogeneity. Instead, researchers are attempting to assess and understand crowdsourced data. Approaches based on statistics are helpful to comprehend trends in crowd-drawing behaviors. This study examines trends in contributors’ first decisions when drawing OpenStreetMap (OSM) buildings. The proposed approach evaluates how important the properties of a point are in determining the first point of building drawings. It classifies the adjacency types of the buildings using a random forest classifier for the properties and aids in inferring drawing trends from the relative impact of each property. To test the approach, detached and attached building groups in Istanbul and Izmir, Turkey, were used. The result had an 83% F-score. In summary, the volunteers tended to choose as first points those further away from the street and building centroid and provided lower point density in the detached buildings than the attached ones. This means that OSM volunteers paid more attention to open spaces when drawing the first points of the detached buildings in the study areas. The study reveals common drawing trends in building-mapping actions.

A Concentric Loop Convolutional Neural Network for Manual Delineation-Level Building Boundary Segmentation From Remote-Sensing Images

To date, accurate building footprint delineation in the surveying, mapping, and geographic information system (GIS) communities has been dependent on human labor. In this article, to address this issue, we propose a concentric loop convolutional neural network (CLP-CNN) method for the automatic segmentation of building boundaries from remote-sensing images. The proposed method consists of three components: 1) a boundary detector to extract coarse polygonal boundaries of individual regions of interest; 2) a concentric loop-shaped convolutional network with bidirectional pairing loss to fine-tune the vertices of the polygons; and 3) a refinement block, which removes redundant vertices and regularizes the boundaries to polygons at the manual delineation level. We also demonstrate that the proposed CLP-CNN method is applicable to other generic objects in natural images. Experiments on two building datasets confirmed that more than 77%/67% of the building polygons predicted by the proposed method are on par with the manual delineation level, representing a significant saving in the labor cost of manual annotation. In generic object boundary delineation tests performed on the Semantic Boundaries Dataset (SBD), the proposed method outperformed the most recent state-of-the-art methods by at least 3.1% in average precision (AP). Furthermore, compared with other vertex matching methods, the learning process of the proposed method converges faster. The source code will be available at <uri xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">http://gpcv.whu.edu.cn/data</uri> .