Building Extraction Research Articles

A Geometric Significance-Aware Deep Mutual Learning Network for Building Extraction from Aerial Images

Knowledge-driven building extraction method exhibits a restricted adaptability scope and is vulnerable to external factors that affect its extraction accuracy. On the other hand, data-driven building extraction method lacks interpretability, heavily relies on extensive training data, and may result in extraction outcomes with building boundary blur issues. The integration of pre-existing knowledge with data-driven learning is essential for the intelligent identification and extraction of buildings from high-resolution aerial images. To overcome the limitations of current deep learning building extraction networks in effectively leveraging prior knowledge of aerial images, a geometric significance-aware deep mutual learning network (GSDMLNet) is proposed. Firstly, the GeoSay algorithm is utilized to derive building geometric significance feature maps as prior knowledge and integrate them into the deep learning network to enhance the targeted extraction of building features. Secondly, a bi-directional guidance attention module (BGAM) is developed to facilitate deep mutual learning between the building feature map and the building geometric significance feature map within the dual-branch network. Furthermore, the deployment of an enhanced flow alignment module (FAM++) is utilized to produce high-resolution, robust semantic feature maps with strong interpretability. Ultimately, a multi-objective loss function is crafted to refine the network’s performance. Experimental results demonstrate that the GSDMLNet excels in building extraction tasks within densely populated and diverse urban areas, reducing misidentification of shadow-obscured regions and color-similar terrains lacking building structural features. This approach effectively ensures the precise acquisition of urban building information in aerial images.

Comparative Analysis of Deep Learning Models for Building Extraction from High-resolution Satellite Imagery

In this research, an approach to extract buildings from Google's satellite imagery was proposed. The performances of various deep learning models (U-Net, RIU-Net, U-Net++, Res-U-Net, and DeepLabV3+) on pre-processed datasets were compared. The models were trained using the similarity metrics of Intersection over Union (IoU) and Dice Similarity Coefficient (DSC). The best-performing models among the segmentation techniques were Res-U-Net and DeepLabV3+. Res-U-Net, an enhanced version of the traditional U-Net model that incorporates residual connections for improved feature propagation, achieved an F1 score of 85.43% when using the RGB dataset. Similarly, DeepLabV3+ also achieved high performance on the Enhanced RGB dataset, obtaining an F1 score of 85.18% after applying pre-processing techniques. This research highlights the significance of color as a dominant feature for accurate building extraction from satellite images. The findings contribute to improved methodologies for building identification, benefiting urban planning, and disaster management applications.

Asymmetric Network Combining CNN and Transformer for Building Extraction from Remote Sensing Images.

The accurate extraction of buildings from remote sensing images is crucial in fields such as 3D urban planning, disaster detection, and military reconnaissance. In recent years, models based on Transformer have performed well in global information processing and contextual relationship modeling, but suffer from high computational costs and insufficient ability to capture local information. In contrast, convolutional neural networks (CNNs) are very effective in extracting local features, but have a limited ability to process global information. In this paper, an asymmetric network (CTANet), which combines the advantages of CNN and Transformer, is proposed to achieve efficient extraction of buildings. Specifically, CTANet employs ConvNeXt as an encoder to extract features and combines it with an efficient bilateral hybrid attention transformer (BHAFormer) which is designed as a decoder. The BHAFormer establishes global dependencies from both texture edge features and background information perspectives to extract buildings more accurately while maintaining a low computational cost. Additionally, the multiscale mixed attention mechanism module (MSM-AMM) is introduced to learn the multiscale semantic information and channel representations of the encoder features to reduce noise interference and compensate for the loss of information in the downsampling process. Experimental results show that the proposed model achieves the best F1-score (86.7%, 95.74%, and 90.52%) and IoU (76.52%, 91.84%, and 82.68%) compared to other state-of-the-art methods on the Massachusetts building dataset, the WHU building dataset, and the Inria aerial image labeling dataset.

Improved Building Extraction from Remotely Sensed Images by Integration of Encode–Decoder and Edge Enhancement Models

Improved Building Extraction from Remotely Sensed Images by Integration of Encode–Decoder and Edge Enhancement Models

A method for extracting buildings from remote sensing images based on 3DJA-UNet3+

Building extraction aims to extract building pixels from remote sensing imagery, which plays a significant role in urban planning, dynamic urban monitoring, and many other applications. UNet3+ is widely applied in building extraction from remote sensing images. However, it still faces issues such as low segmentation accuracy, imprecise boundary delineation, and the complexity of network models. Therefore, based on the UNet3+ model, this paper proposes a 3D Joint Attention (3DJA) module that effectively enhances the correlation between local and global features, obtaining more accurate object semantic information and enhancing feature representation. The 3DJA module models semantic interdependence in the vertical and horizontal dimensions to obtain feature map spatial encoding information, as well as in the channel dimensions to increase the correlation between dependent channel graphs. In addition, a bottleneck module is constructed to reduce the number of network parameters and improve model training efficiency. Many experiments are conducted on publicly accessible WHU,INRIA and Massachusetts building dataset, and the benchmarks, BOMSC-Net, CVNet, SCA-Net, SPCL-Net, ACMFNet, MFCF-Net models are selected for comparison with the 3DJA-UNet3+ model proposed in this paper. The experimental results show that 3DJA-UNet3+ achieves competitive results in three evaluation indicators: overall accuracy, mean intersection over union, and F1-score. The code will be available at https://github.com/EnjiLi/3DJA-UNet3Plus.

FwSVM-Net: A novel deep learning-based automatic building extraction from aerial images

The building layer, one of the fundamental elements of mapping, features various layouts and architectures and is used in many mapping activities, such as urban planning, property management, illegal building detection, and disaster investigation. Therefore, accessing building information from remotely sensed images in a fast and automated manner is a popular topic. This paper aims to design a new Convolutional Neural Network (CNN) architecture called FwSVM-Net (Fast with Support Vector Machine Network) for building extraction. The FwSVM-Net, with 16 convolutional layers, employs a fusion mechanism to reduce the semantic gap between the encoder and decoder sections. In the architecture's classification layer, a Support Vector Machine (SVM) is used to increase segmentation accuracy while reducing parameter density. The Massachusetts Building Dataset was used for the automatic building extraction from aerial images. The performance of the FwSVM-Net on the test data was calculated at 97 %, 91 %, 87 %, 89 %, and 86 % for Overall Accuracy (OA), Precision (Pr), Recall (Rc), F1-Score (F1), and Intersection over Union (IoU), respectively. FwSVM-Net shows approximately ±2 % similarity in accuracy with the U-Net architecture, which is trained and evaluated with the same dataset. Despite this, the fact that the FwSVM-Net completed the training process approximately twice as fast as the U-Net architecture provides a significant time advantage. As a result, it is predicted that the FwSVM-Net will offer speed and convenience in terms of usability because of its performance, which is equivalent to the performance of the U-Net architecture for segmentation, and its time advantage.

Semi-Supervised Building Extraction with Optical Flow Correction Based on Satellite Video Data in a Tsunami-Induced Disaster Scene.

Data and reports indicate an increasing frequency and intensity of natural disasters worldwide. Buildings play a crucial role in disaster responses and damage assessments, aiding in planning rescue efforts and evaluating losses. Despite advances in applying deep learning to building extraction, challenges remain in handling complex natural disaster scenes and reducing reliance on labeled datasets. Recent advances in satellite video are opening a new avenue for efficient and accurate building extraction research. By thoroughly mining the characteristics of disaster video data, this work provides a new semantic segmentation model for accurate and efficient building extraction based on a limited number of training data, which consists of two parts: the prediction module and the automatic correction module. The prediction module, based on a base encoder-decoder structure, initially extracts buildings using a limited amount of training data that are obtained instantly. Then, the automatic correction module takes the output of the prediction module as input, constructs a criterion for identifying pixels with erroneous semantic information, and uses optical flow values to extract the accurate corresponding semantic information on the corrected frame. The experimental results demonstrate that the proposed method outperforms other methods in accuracy and computational complexity in complicated natural disaster scenes.

Building extraction from remote sensing imagery: advanced squeeze-and-excitation residual network based methodology

Extracting buildings from remote sensing imagery (RSI) is an essential task in a wide range of applications, such as urban and monitoring. Deep learning has emerged as a powerful tool for this purpose, and in this research, we propose an advanced building extraction method based on SE-ResNet18 and SE-ResNet34 architectures. These models were selected through a rigorous comparative analysis of various deep learning models, including variations of residual networks (ResNet), squeeze-and-excitation residual networks (SE-ResNet), and visual geometry group (VGG), for their high performance in all metrics and their computational efficiency. Our proposed methodology outperformed all other models under consideration by a significant margin, demonstrating its robustness and efficiency. It achieved superior results with less computational effort and time, a testament to its potential as a powerful tool for semantic segmentation tasks in remote sensing applications. An extensive comparative evaluation involving a wide range of state-of-the-art works further validated our method’s effectiveness. Our method achieved an unparalleled intersection over union (IoU) score of 88.51%, indicative of its exceptional accuracy in identifying and segmenting buildings within the Wuhan University (WHU) building dataset. The overall performance of our method, which offers an excellent balance between high performance and computational efficiency, makes it a compelling choice for researchers and practitioners in the field.

Unsupervised domain adaptive building semantic segmentation network by edge-enhanced contrastive learning

Unsupervised domain adaptation (UDA) is a weakly supervised learning technique that classifies images in the target domain when the source domain has labeled samples, and the target domain has unlabeled samples. Due to the complexity of imaging conditions and the content of remote sensing images, the use of UDA to accurately extract artificial features such as buildings from high-spatial-resolution (HSR) imagery is still challenging. In this study, we propose a new UDA method for building extraction, the contrastive domain adaptation network (CDANet), by utilizing adversarial learning and contrastive learning techniques. CDANet consists of a single multitask generator and dual discriminators. The generator employs a region and edge dual-branch structure that strengthens its edge extraction ability and is beneficial for the extraction of small and densely distributed buildings. The dual discriminators receive the region and edge prediction outputs and achieve multilevel adversarial learning. During adversarial training processing, CDANet aligns the cross-domain of similar pixel features in the embedding space by constructing the regional pixelwise contrastive loss. A self-training (ST) strategy based on pseudolabel generation is further utilized to address the target intradomain discrepancy. Comprehensive experiments are conducted to validate CDANet on three publicly accessible datasets, namely the WHU, Austin, and Massachusetts. Ablation experiments show that the generator network structure, contrastive loss and ST strategy all improve the building extraction accuracy. Method comparisons validate that CDANet achieves superior performance to several state-of-the-art methods, including AdaptSegNet, AdvEnt, IntraDA, FDANet and ADRS, in terms of F1 score and mIoU.

Automatic Building Extraction from Multispectral LiDAR Using Novel 3D Spatial Indices and Deep Learning Point CNN

Automatic Building Extraction from Multispectral LiDAR Using Novel 3D Spatial Indices and Deep Learning Point CNN

Enhanced large-scale building extraction evaluation: developing a two-level framework using proxy data and building matching

ABSTRACT Deep learning-based building extraction methods have widespread applications in diverse fields. However, the evaluation of large-scale extraction results remains challenging, due to traditional evaluation metrics rely on manually created ground-truth samples and the lack of comprehensive reference-building data for developing countries. To address these problems, we proposed a two-level framework for evaluating large-scale footprint extraction. First, we utilised global open-source population and land use data as the proxy data, to assess grid-level completeness for the areas with insufficient reference data. Second, we introduced an improved two-way area-overlapping method to match the extracted footprints with the reference buildings, thereby enabling a comprehensive evaluation of the study region. Tested in Hyogo Prefecture and Numazu City, Japan, the results demonstrated a 2.6-% improvement in grid classification accuracy and an increase of 0.53 in the completeness correlation, compared with the results obtained using a single proxy indicator. Moreover, the optimised matching method achieved an outstanding semantic matching accuracy of 99%, with high efficiency and robustness in multi-scale matching. Therefore, the proposed approach can effectively evaluate large-scale footprint extraction results and interpret their semantic relationship with actual buildings, applicable globally regardless of the availability of reference building datasets.

EUNet: Edge-UNet for Accurate Building Extraction and Edge Emphasis in Gaofen-7 Images

Deep learning is currently the mainstream approach for building extraction tasks in remote-sensing imagery, capable of automatically learning features of buildings in imagery and yielding satisfactory extraction results. However, due to the diverse sizes, irregular layouts, and complex spatial relationships of buildings, extracted buildings often suffer from incompleteness and boundary issues. Gaofen-7 (GF-7), as a high-resolution stereo mapping satellite, provides well-rectified images from its rear-view imagery, which helps mitigate occlusions in highly varied terrain, thereby offering rich information for building extraction. To improve the integrity of the edges of the building extraction results, this paper proposes a dual-task network (Edge-UNet, EUnet) based on UNet, incorporating an edge extraction branch to emphasize edge information while predicting building targets. We evaluate this method using a self-made GF-7 Building Dataset, the Wuhan University (WHU) Building Dataset, and the Massachusetts Buildings Dataset. Comparative analysis with other mainstream semantic segmentation networks reveals significantly higher F1 scores for the extraction results of our method. Our method exhibits superior completeness and accuracy in building edge extraction compared to unmodified algorithms, demonstrating robust performance.

China's first sub-meter building footprints derived by deep learning

The high spatial resolution building footprints are crucial for understanding urban development and its associated applications. However, up to now, the sub-meter-level building footprint data of China is still lacking. The challenges arise from two aspects: 1) the number of training samples is inadequate for large-scale building extraction. 2) the accuracy and efficiency of current models are insufficient to conduct large-scale building extraction. Therefore, we propose a framework for large-scale building extraction in this study, including semi-automated sample generation, building extraction model, model training, and post-processing. Specifically, the main technical contributions include: 1) BldgNet (Building Extraction Network) is proposed, including the Large Window Attention, Edge Attention, and Distribution Alignment Module with consideration of spatial contextual information, to address the challenge of the multi-scale building extraction, building boundary delineation, and class imbalance, respectively; 2) a semi-supervised training approach is proposed for large-scale building extraction, leveraging the incomplete information from OpenStreetMap (OSM) to enhance the diversity of building samples and the robustness of the model. Meanwhile, we created an open-source Global Building Dataset (GBD) comprising approximately 800,000 high-resolution (0.25 m) samples. This dataset incorporates diverse building styles worldwide, offering support for global building extraction. Based on the constructed sample set and the proposed deep net, we generated China's first sub-meter (0.5 m) building footprint dataset (CBF). Through testing on 750,000 buildings from 350 cities, the overall F1 score for CBF reached 83.71%. Finally, we validated that the proposed building extraction model can achieve satisfactory results compared to existing representative deep networks. GBD and CBF datasets can be publicly available and downloadable via https://zenodo.org/doi/10.5281/zenodo.10043351.

Assessing the contribution of RGB VIs in improving building extraction from RGB-UAV images

Buildings are a fundamental component of the built environment, and accurate information regarding their size, location, and distribution is vital for various purposes. The ever-increasing capabilities of unmanned aerial vehicles (UAVs) have sparked an interest in exploring various techniques to delineate buildings from the very high-resolution images obtained from UAV photogrammetry. However, the limited spectral information in UAV images, particularly the number of bands, can hinder the differentiation between various materials and objects. This setback can affect the ability to distinguish between different materials and objects. To address this limitation, vegetative ındices (VIs) have been employed to enhance the spectral strength of UAV orthophotos, thereby improving building classification. The objective of this study is to evaluate the contribution of four specific VIs: the green leaf index (GLI), red-green-blue vegetation index (RGBVI), visual atmospherically resistant index (VARI), and triangular greenness index (TGI). The significance of this contribution lies in assessing the potential of each VI to enhance building classification. The approach utilized the geographic object-based image analysis (GeoBIA) approach and a random forest classifier. To achieve this aim, five datasets were created, with each dataset comprising the RGB-UAV image and a corresponding RGB VI. The experimental results on the test dataset and a post-classification assessment indicated a general improvement in the classification when the VIs were added to the RGB orthophoto.

Target-based building extraction from high-resolution RGB imagery using GEOBIA framework and tabular deep learning model

Extracting building information from remotely sensed images is essential in various geographic and environmental applications. Specifically, there is a growing demand for distinguishing buildings based on roof colors from low-cost but high-resolution RGB images. However, this task presents significant challenges due to the variability in building color, size, and style, as well as their resemblance to nonbuilding objects. The existing GEographic Object-Based Image Analysis (GEOBIA) approaches often struggle with misclassification, mainly because they do not effectively leverage spatial information during the classification stage. Tabular Deep Learning (DL) models have emerged as promising alternatives to shallow classifiers, demonstrating improved performance in recent years. However, their effectiveness in extracting buildings based on roof color has not been thoroughly examined. To address these challenges and improve the accuracy of building extraction, this study proposed a target-based model that incorporates a sequential extraction process using tabular DL models such as Gated Additive Tree Ensemble (GATE), Neural Oblivious Decision Ensemble (NODE). The proposed method adopted a hybrid segmentation approach to effectively segment the image and extract each segment's spectral, textural, geometrical, and contextual features. Non-building objects were systematically eliminated based on their representation within the image. Subsequently, buildings with varying roof colors—brown, black, and white—were extracted. Spatial information was then meticulously integrated into the classification process. The proposed methodology was tested on a UAV image with a spatial resolution of 0.20 m and three visible bands—red, green, and blue—captured in Kingston, Ontario, Canada. This enhanced approach significantly improved precision and overall classification accuracy, demonstrating the effectiveness of incorporating spatial data into the extraction process. This research introduced a framework for incorporating spatial information into tabular models that would serve as a foundation for further improving building extraction accuracy in future studies.

INTEGRATION OF RESUNET AND YOLO ALGORITHMS INTO A UNIFIED MODEL FOR OBJECTS DETECTION

<p>Automatic extraction of footprints of buildings from orthophotos is a challenge in the field of remote sensing data processing. The combination of image classification and object localization tasks in this research aims to develop a new model based on ResUNet and the YOLO algorithm. By applying the proposed model and publicly available data, a high level of building extraction success of 89% is achieved. Although there is potential to improve the results by introducing other types of data, the integration of these models represents a significant step towards improving the technology of automatic extraction of buildings from orthophotos.</p>

Pre-trained regional models for extracting buildings from high resolution satellite imagery to support public health initiatives

Up-to-date building locations provide a pivotal layer of information for public health initiatives and emergency responses to identify and reach a target population. Using pre-trained deep learning models to extract buildings from high-resolution imagery reduces the time and resources required but often under-perform in new locations with landscapes and building types different from that used for training the models. This study aims to develop and evaluate deep learning models for extracting buildings from high-resolution satellite imagery for a region of interest. The main objectives are to use openly available codebase, base models, and training datasets from a region to build models, to use the pre-trained regional models to extract buildings from other locations in the region, and to assess the accuracies of the models. The regional models are also fine-tuned with data from a new location to build and evaluate local models. Two regional models for Bangladesh outperformed a baseline model for low- and middle-income countries, with a 30% increase in mean accuracy (F1-scores), in three test sites. Among the regional models, a model trained with data exclusively from the region performed better (mean F1-score: 0.57) than a model built by fine-tuning the baseline model with regional training data (mean F1-score: 0.45). Local models based on the regional models also outperformed other local models. This study demonstrates the potential of pre-trained regional models in expediting building extraction, but points out the challenges associated with limited training data.

Contour Extraction of UAV Point Cloud Based on Neighborhood Geometric Features of Multi-Level Growth Plane

The extraction of UAV building point cloud contour points is the basis for the expression of a three-dimensional lightweight building outline. Previous unmanned aerial vehicle (UAV) building point cloud contour extraction methods have mainly focused on the expression of the roof contour, but did not extract the wall contour. In view of this, an algorithm based on the geometric features of the neighborhood points of the region-growing clustering fusion surface is proposed to extract the boundary points of the UAV building point cloud. Firstly, the region growth plane is fused to obtain a more accurate segmentation plane. Then, the neighboring points are projected onto the neighborhood plane and a vector between the object point and neighborhood point is constructed. Finally, the azimuth of each vector is calculated, and the boundary points of each segmented plane are extracted according to the difference in adjacent azimuths. Experiment results show that the best boundary points can be extracted when the number of adjacent points is 24 and the difference in adjacent azimuths is 120. The proposed method is superior to other methods in the contour extraction of UAV buildings point clouds. Moreover, it can extract not only the building roof contour points, but also the wall contour points, including the window contour points.

Building extraction in urban and rural areas with aerial and LiDAR DSM

Building extraction in urban and rural areas with aerial and LiDAR DSM

Deep Learning based Semantic Segmentation for Buildings Detection from Remote Sensing Images

Building extraction from remote sensing images is the process of automatically identifying and extracting the boundaries of buildings from high-resolution aerial or satellite images. The extracted building footprints can be used for a variety of applications, such as urban planning, disaster management, city development, land management, environmental monitoring, and 3D modeling. The results of building extraction from remote sensing images depend on several factors, such as the quality and resolution of the image and the choice of algorithm.The process of building extraction from remote sensing images typically involves a series of steps, including image pre-processing, feature extraction, and classification. Building extraction from remote sensing images can be challenging due to factors such as varying building sizes and shapes, shadows, and occlusions. However, recent advances in deep learning and computer vision techniques have led to significant improvements in the accuracy and efficiency of building extraction methods. This research presents a deep learning semantic segmentation architecture-based model for developing building detection from high resolution remote sensing images. The open-source Massachusetts dataset is used to train the suggested UNet architecture. The model is optimized using the RMSProp algorithm with a learning rate of 0.0001 for 100 epochs. After 1.52 hours of training on Google Colab the model achieved an 83.55% F1 score, which indicates strong precision and recall.