A Stacking Ensemble Deep Learning Model for Building Extraction from Remote Sensing Images

Abstract

Automatically extracting buildings from remote sensing images with deep learning is of great significance to urban planning, disaster prevention, change detection, and other applications. Various deep learning models have been proposed to extract building information, showing both strengths and weaknesses in capturing the complex spectral and spatial characteristics of buildings in remote sensing images. To integrate the strengths of individual models and obtain fine-scale spatial and spectral building information, this study proposed a stacking ensemble deep learning model. First, an optimization method for the prediction results of the basic model is proposed based on fully connected conditional random fields (CRFs). On this basis, a stacking ensemble model (SENet) based on a sparse autoencoder integrating U-NET, SegNet, and FCN-8s models is proposed to combine the features of the optimized basic model prediction results. Utilizing several cities in Hebei Province, China as a case study, a building dataset containing attribute labels is established to assess the performance of the proposed model. The proposed SENet is compared with three individual models (U-NET, SegNet and FCN-8s), and the results show that the accuracy of SENet is 0.954, approximately 6.7%, 6.1%, and 9.8% higher than U-NET, SegNet, and FCN-8s models, respectively. The identification of building features, including colors, sizes, shapes, and shadows, is also evaluated, showing that the accuracy, recall, F1 score, and intersection over union (IoU) of the SENet model are higher than those of the three individual models. This suggests that the proposed ensemble model can effectively depict the different features of buildings and provides an alternative approach to building extraction with higher accuracy.

Mapping Vegetation Types by Different Fully Convolutional Neural Network Structures with Inadequate Training Labels in Complex Landscape Urban Areas

“…In contrast to U-Net, UNS showed less "salt-and-pepper" (Figure 13D). Therefore, combining deep learning and ensemble learning, UNS was able to enhance the generalization and robustness of the model by combining deep features extracted by different networks and effectively extracting information on urban vegetation over large areas [44].…”

Diagnosis Myocardial Infarction Based on Stacking Ensemble of Convolutional Neural Network

“…Ensemble techniques are approaches that combine different learning algorithms or models to create a single, ideal prediction model that performs better than the base learners considered separately [13][14][15]. The ensemble learning method involves three different methods, the difference of which is centered on how the inputs are combined to obtain distinct outputs, namely stacking, boosting, and bagging [16].…”

Overview of the Special Issue on Applications of Remote Sensing Imagery for Urban Areas

“…This Special Issue (SI) aims to invite recent advances in the applications of RS imagery for urban areas, and 17 papers in total were selected and published. Among them, 12 papers emphasize the novel urban application algorithms based on RS imageries, such as urban attribute mapping, building extraction, classification, change detection, and so on [1][2][3][4][5][6][7][8][9][10][11][12], and 5 papers directly employed RS imageries to analyze the environmental variations and urban expansion in typical cities, such as urban heat island, air pollution, lightning, and so on [13][14][15][16][17].…”

“…RS imageries provide new opportunities to extract the urban building information and detect its changes, and thus there are four papers focused on this issue [1][2][3][4]. Cao et al [1] proposed a stacking ensemble deep learning model (SENet) to obtain fine-scale spatial and spectral building information, based on a sparse autoencoder integrating U-NET, SegNet, and FCN-8s models. The model was assessed by a building dataset in Hebei Province, China, and the results indicate that its accuracy is significantly improved compared to all three models.…”

“…In summary, this SI provides an enhanced understanding of applications of RS imagery for urban areas. New methodologies are presented for extracting the building information, modeling urban attributes and population, and detecting the urban flooding using remote sensing imageries [1][2][3][4]10,12]. Novel models for RS classification, target and change detection are also included, which can significantly support further urban applications [5][6][7][8][9].…”