Search

Accurate urban land cover information is crucial for effective urban planning and management. While convolutional neural networks (CNNs) demonstrate superior feature learning and prediction capabilities using image-level annotations, the inherent mixed-category nature of input image patches leads to classification errors along object boundaries. Fully convolutional neural networks (FCNs) excel at pixel-wise fine segmentation, making them less susceptible to heterogeneous content, but they require fully annotated dense image patches, which may not be readily available in real-world scenarios. This paper proposes an object-based semi-supervised spatial attention residual UNet (OS-ARU) model. First, multiscale segmentation is performed to obtain segments from a remote sensing image, and segments containing sample points are assigned the categories of the corresponding points, which are used to train the model. Then, the trained model predicts class probabilities for all segments. Each unlabeled segment's probability distribution is compared against those of labeled segments for similarity matching under a threshold constraint. Through label propagation, pseudo-labels are assigned to unlabeled segments exhibiting high similarity to labeled ones. Finally, the model is retrained using the augmented training set incorporating the pseudo-labeled segments. Comprehensive experiments on aerial image benchmarks for Vaihingen and Potsdam demonstrate that the proposed OS-ARU achieves higher classification accuracy than state-of-the-art models, including OCNN, 2OCNN, and standard OS-U, reaching an overall accuracy (OA) of 87.83% and 86.71%, respectively. The performance improvements over the baseline methods are statistically significant according to the Wilcoxon Signed-Rank Test. Despite using significantly fewer sparse annotations, this semi-supervised approach still achieves comparable accuracy to the same model under full supervision. The proposed method thus makes a step forward in substantially alleviating the heavy sampling burden of FCNs (densely sampled deep learning models) to effectively handle the complex issue of land cover information identification and classification. [ABSTRACT FROM AUTHOR]

Transformer models have achieved great results in the field of computer vision over the past 2 years, drawing attention from within the field of remote sensing. However, there are still relatively few studies on this model in the field of remote sensing. Which method is more suitable for remote-sensing segmentation? In particular, how do different transformer models perform in the face of high-spatial resolution and the multispectral resolution of remote-sensing images? To explore these questions, this paper presents a comprehensive comparative analysis of three mainstream transformer models, including the segmentation transformer (SETRnet), SwinUnet, and TransUnet, by evaluating three aspects: a visual analysis of feature-segmentation results, accuracy, and training time. The experimental results show that the transformer structure has obvious advantages for the feature-extraction ability of large-scale remote-sensing data sets and ground objects, but the segmentation performance of different transfer structures in different scales of remote-sensing data sets is also very different. SwinUnet exhibits better global semantic interaction and pixel-level segmentation prediction on the large-scale Potsdam data set, and the SwinUnet model has the highest accuracy metrics for KAPPA, MIoU, and OA in the Potsdam data set, at 76.47%, 63.62%, and 85.01%, respectively. TransUnet has better segmentation results in the small-scale Vaihingen data set, and the three accuracy metrics of KAPPA, MIoU, and OA are the highest, at 80.54%, 56.25%, and 85.55%, respectively. TransUnet is better able to handle the edges and details of feature segmentation thanks to the network structure together built by its transformer and convolutional neural networks (CNNs). Therefore, TransUnet segmentation accuracy is higher when using a small-scale Vaihingen data set. Compared with SwinUnet and TransUnet, the segmentation performance of SETRnet in different scales of remote-sensing data sets is not ideal, so SETRnet is not suitable for the research task of remote-sensing image segmentation. In addition, this paper discusses the reasons for the performance differences between transformer models and discusses the differences between transformer models and CNN. This study further promotes the application of transformer models in remote-sensing image segmentation, improves the understanding of transformer models, and helps relevant researchers to select a more appropriate transformer model or model improvement method for remote-sensing image segmentation. [ABSTRACT FROM AUTHOR]

The semantic segmentation of remote sensing images is a significant research direction in digital image processing. The complex background environment, irregular size and shape of objects, and similar appearance of different categories of remote sensing images have brought great challenges to remote sensing image segmentation tasks. Traditional convolutional-neural-network-based models often ignore spatial information in the feature extraction stage and pay less attention to global context information. However, spatial context information is important in complex remote sensing images, which means that the segmentation effect of traditional models needs to be improved. In addition, neural networks with a superior segmentation performance often suffer from the problem of high computational resource consumption. To address the above issues, this paper proposes a combination model of a modified multiscale deformable convolutional neural network (mmsDCNN) and dense conditional random field (DenseCRF). Firstly, we designed a lightweight multiscale deformable convolutional network (mmsDCNN) with a large receptive field to generate a preliminary prediction probability map at each pixel. The output of the mmsDCNN model is a coarse segmentation result map, which has the same size as the input image. In addition, the preliminary segmentation result map contains rich multiscale features. Then, the multi-level DenseCRF model based on the superpixel level and the pixel level is proposed, which can make full use of the context information of the image at different levels and further optimize the rough segmentation result of mmsDCNN. To be specific, we converted the pixel-level preliminary probability map into a superpixel-level predicted probability map according to the simple linear iterative clustering (SILC) algorithm and defined the potential function of the DenseCRF model based on this. Furthermore, we added the pixel-level potential function constraint term to the superpixel-based Gaussian potential function to obtain a combined Gaussian potential function, which enabled our model to consider the features of various scales and prevent poor superpixel segmentation results from affecting the final result. To restore the contour of the object more clearly, we utilized the Sketch token edge detection algorithm to extract the edge contour features of the image and fused them into the potential function of the DenseCRF model. Finally, extensive experiments on the Potsdam and Vaihingen datasets demonstrated that the proposed model exhibited significant advantages compared to the current state-of-the-art models. [ABSTRACT FROM AUTHOR]

Single-view height estimation and semantic segmentation have received increasing attention in recent years and play an important role in the photogrammetry and remote sensing communities. The height information and semantic information of images are correlated, and some recent works have shown that multi-task learning methods can achieve complementation of task-related features and improve the prediction results of the multiple tasks. Although much progress has been made in recent works, how to effectively extract and fuse height features and semantic features is still an open issue. In this paper, a self- and cross-enhancement network (SCE-Net) is proposed to jointly perform height estimation and semantic segmentation on single aerial images. A feature separation–fusion module is constructed to effectively separate and fuse height features and semantic features based on an attention mechanism for feature representation enhancement across tasks. In addition, a height-guided feature distance loss and a semantic-guided feature distance loss are designed based on deep metric learning to achieve task-aware feature representation enhancement. Extensive experiments are conducted on the Vaihingen dataset and the Potsdam dataset to verify the effectiveness of the proposed method. The experimental results demonstrate that the proposed SCE-Net could outperform the state-of-the-art methods and achieve better performance in both height estimation and semantic segmentation. [ABSTRACT FROM AUTHOR]

Semantic segmentation of remote sensing imagery (RSI) has obtained great success with the development of deep convolutional neural networks (DCNNs). However, most of the existing algorithms focus on designing end-to-end DCNNs, but neglecting to consider the difficulty of segmentation in imbalance categories, especially for minority categories in RSI, which limits the performance of RSI semantic segmentation. In this paper, a novel edge guided context aggregation network (EGCAN) is proposed for the semantic segmentation of RSI. The Unet is employed as backbone. Meanwhile, an edge guided context aggregation branch and minority categories extraction branch are designed for a comprehensive enhancement of semantic modeling. Specifically, the edge guided context aggregation branch is proposed to promote entire semantic comprehension of RSI and further emphasize the representation of edge information, which consists of three modules: edge extraction module (EEM), dual expectation maximization attention module (DEMA), and edge guided module (EGM). EEM is created primarily for accurate edge tracking. According to that, DEMA aggregates global contextual features with different scales and the edge features along spatial and channel dimensions. Subsequently, EGM cascades the aggregated features into the decoder process to capture long-range dependencies and further emphasize the error-prone pixels in the edge region to acquire better semantic labels. Besides this, the exploited minority categories extraction branch is presented to acquire rich multi-scale contextual information through an elaborate hybrid spatial pyramid pooling module (HSPP) to distinguish categories taking a small percentage and background. On the Tianzhi Cup dataset, the proposed algorithm EGCAN achieved an overall accuracy of 84.1% and an average cross-merge ratio of 68.1%, with an accuracy improvement of 0.4% and 1.3% respectively compared to the classical Deeplabv3+ model. Extensive experimental results on the dataset released in ISPRS Vaihingen and Potsdam benchmarks also demonstrate the effectiveness of the proposed EGCAN over other state-of-the-art approaches. [ABSTRACT FROM AUTHOR]

Foremost deep neural network models trained in natural scenes cannot transfer and apply to remote sensing image semantic segmentation well. Studies have shown that fine-tuning methods containing model fusion can alleviate this dilemma. In this paper, we provide an approach used to improve U-Net and propose an end-to-end deep convolutional neural network (DCNN) combining the superiorities of DenseNet, U-Net, dilated convolution, and DeconvNet. We evaluated the proposed method and model on the Potsdam orthophoto data set. Compared with U-Net, our approach increases the PA, mPA, and mIoU evaluation indexes by 11.1%, 14.0%, and 13.5%, respectively; the segmentation speed increases by approximately 1.18 times and the number of parameters is 59.0% that of U-Net. The experiments demonstrate that for the semantic segmentation of high-resolution remote sensing images, using the combined dilated convolutions as the primary feature extractor, using the transposed convolution to restore the size of the feature maps, and reducing the number of layers is an effective method to improve the comprehensive performance of U-Net. This research enriches the models based on DCNNs and the modes of using DCNNs in a specific scene. [ABSTRACT FROM AUTHOR]

Despite the significant advances noted in semantic segmentation of aerial imagery, a considerable limitation is blocking its adoption in real cases. If we test a segmentation model on a new area that is not included in its initial training set, accuracy will decrease remarkably. This is caused by the domain shift between the new targeted domain and the source domain used to train the model. In this paper, we addressed this challenge and proposed a new algorithm that uses Generative Adversarial Networks (GAN) architecture to minimize the domain shift and increase the ability of the model to work on new targeted domains. The proposed GAN architecture contains two GAN networks. The first GAN network converts the chosen image from the target domain into a semantic label. The second GAN network converts this generated semantic label into an image that belongs to the source domain but conserves the semantic map of the target image. This resulting image will be used by the semantic segmentation model to generate a better semantic label of the first chosen image. Our algorithm is tested on the ISPRS semantic segmentation dataset and improved the global accuracy by a margin up to 24% when passing from Potsdam domain to Vaihingen domain. This margin can be increased by addition of other labeled data from the target domain. To minimize the cost of supervision in the translation process, we proposed a methodology to use these labeled data efficiently. [ABSTRACT FROM AUTHOR]

With developments in deep learning, semantic segmentation of remote sensing images has made great progress. Currently, mainstream methods are based on convolutional neural networks (CNNs) or vision transformers. However, these methods are not very effective in extracting features from remote sensing images, which are usually of high resolution with plenty of detail. Operations including downsampling will cause the loss of such features. To address this problem, we propose a novel module called Hierarchical Wavelet Feature Enhancement (WFE). The WFE module involves three sequential steps: (1) performing multi-scale decomposition of an input image based on the discrete wavelet transform; (2) enhancing the high-frequency sub-bands of the input image; and (3) feeding them back to the corresponding layers of the network. Our module can be easily integrated into various existing CNNs and transformers, and does not require additional pre-training. We conducted experiments on the ISPRS Potsdam and ISPRS Vaihingen datasets, with results showing that our method improves the benchmarks of CNNs and transformers while performing little additional computation. [ABSTRACT FROM AUTHOR]

This paper presents a novel multi-pooling architecture generated by combining the advantages of wavelet and max-pooling operations in convolutional neural networks (CNNs), focusing on semantic segmentation tasks. CNNs often use pooling to reduce the number of parameters, improve invariance to certain distortions, and enlarge the receptive field. However, pooling can cause information loss and thus is detrimental to further operations such as feature extraction and analysis. This problem is particularly critical for semantic segmentation, where each pixel of an image is assigned to a specific class to divide the image into disjoint regions of interest. To address this problem, pooling strategies based on wavelets-operations have been proposed with the promise to achieve a better trade-off between receptive field size and computational efficiency. Previous works have confirmed the superiority of wavelet pooling over the traditional one in semantic segmentation tasks. However, we have observed in our computational experiments that the expressive gains reported from the use of wavelet pooling in other segmentation tasks were not observed in the scope of aerial imagery due to imprecision in the segmentation of image details. The combination of wavelet pooling and max-pooling, a solution not yet reported in the literature, can address that issue. Such gap observed in the pooling area motivated the two proposals that are the main contributions of this paper: (a) A new multi-pooling strategy combining wavelet and traditional pooling in a new network structure suitable for aerial image segmentation tasks; (b) Two-stream architectures using the traditional max-pooling and wavelet pooling as streams. These proposals were implemented using the Segnet, a known architecture for semantic segmentation. The computational experiments, based on the IRRG images from the Potsdam and Vaihingen data sets, demonstrated that the proposed architectures surpassed the original Segnet architecture's performance with results comparable to state-of-the-art approaches. [Display omitted] • A new multi-pooling strategy combining wavelet and traditional pooling. • A new version of Segnet, named MPSegnet, using the new multi-pooling strategy. • A two-stream architecture that combines the Segnet network with MPSegnet. • The multi-pooling scheme efficiency was verified using complexity and visual analysis. • Experiments showing that our methods are comparable to the state-of-the-art. [ABSTRACT FROM AUTHOR]

The semantic segmentation of remote sensing images requires distinguishing local regions of different classes and exploiting a uniform global representation of the same-class instances. Such requirements make it necessary for the segmentation methods to extract discriminative local features between different classes and to explore representative features for all instances of a given class. While common deep convolutional neural networks (DCNNs) can effectively focus on local features, they are limited by their receptive field to obtain consistent global information. In this paper, we propose a memory-augmented transformer (MAT) to effectively model both the local and global information. The feature extraction pipeline of the MAT is split into a memory-based global relationship guidance module and a local feature extraction module. The local feature extraction module mainly consists of a transformer, which is used to extract features from the input images. The global relationship guidance module maintains a memory bank for the consistent encoding of the global information. Global guidance is performed by memory interaction. Bidirectional information flow between the global and local branches is conducted by a memory-query module, as well as a memory-update module, respectively. Experiment results on the ISPRS Potsdam and ISPRS Vaihingen datasets demonstrated that our method can perform competitively with state-of-the-art methods. [ABSTRACT FROM AUTHOR]

Recovering height information from a single aerial image is a key problem in the fields of computer vision and remote sensing. At present, supervised learning methods have achieved impressive results, but, due to domain bias, the trained model cannot be directly applied to a new scene. In this paper, we propose a novel semi-supervised framework, StyHighNet, for accurately estimating the height of a single aerial image in a new city that requires only a small number of labeled data. The core is to transfer multi-source images to a unified style, making the unlabeled data provide the appearance distribution as additional supervision signals. The framework mainly contains three sub-networks: (1) the style transferring sub-network maps multi-source images into unified style distribution maps (USDMs); (2) the height regression sub-network, with the function of predicting the height maps from USDMs; and (3) the style discrimination sub-network, used to distinguish the sources of USDMs. Among them, the style transferring sub-network shoulders dual responsibilities: On the one hand, it needs to compute USDMs with obvious characteristics, so that the height regression sub-network can accurately estimate the height maps. On the other hand, it is necessary that the USDMs have consistent distribution to confuse the style discrimination sub-network, so as to achieve the goal of domain adaptation. Unlike previous methods, our style distribution function is learned unsupervised, thus it is of greater flexibility and better accuracy. Furthermore, when the style discrimination sub-network is shielded, this framework can also be used for supervised learning. We performed qualitatively and quantitative evaluations on two sets of public data, Vaihingen and Potsdam. Experiments show that the framework achieved superior performance in both supervised and semi-supervised learning modes. [ABSTRACT FROM AUTHOR]

Semantic segmentation is a significant method in remote sensing image (RSIs) processing and has been widely used in various applications. Conventional convolutional neural network (CNN)-based semantic segmentation methods are likely to lose the spatial information in the feature extraction stage and usually pay little attention to global context information. Moreover, the imbalance of category scale and uncertain boundary information meanwhile exists in RSIs, which also brings a challenging problem to the semantic segmentation task. To overcome these problems, a high-resolution context extraction network (HRCNet) based on a high-resolution network (HRNet) is proposed in this paper. In this approach, the HRNet structure is adopted to keep the spatial information. Moreover, the light-weight dual attention (LDA) module is designed to obtain global context information in the feature extraction stage and the feature enhancement feature pyramid (FEFP) structure is promoted and employed to fuse the contextual information of different scales. In addition, to achieve the boundary information, we design the boundary aware (BA) module combined with the boundary aware loss (BAloss) function. The experimental results evaluated on Potsdam and Vaihingen datasets show that the proposed approach can significantly improve the boundary and segmentation performance up to 92.0% and 92.3% on overall accuracy scores, respectively. As a consequence, it is envisaged that the proposed HRCNet model will be an advantage in remote sensing images segmentation. [ABSTRACT FROM AUTHOR]

Height map estimation from a single aerial image plays a crucial role in localization, mapping, and 3D object detection. Deep convolutional neural networks have been used to predict height information from single-view remote sensing images, but these methods rely on large volumes of training data and often overlook geometric features present in orthographic images. To address these issues, this study proposes a gradient-based self-supervised learning network with momentum contrastive loss to extract geometric information from non-labeled images in the pretraining stage. Additionally, novel local implicit constraint layers are used at multiple decoding stages in the proposed supervised network to refine high-resolution features in height estimation. The structural-aware loss is also applied to improve the robustness of the network to positional shift and minor structural changes along the boundary area. Experimental evaluation on the ISPRS benchmark datasets shows that the proposed method outperforms other baseline networks, with minimum MAE and RMSE of 0.116 and 0.289 for the Vaihingen dataset and 0.077 and 0.481 for the Potsdam dataset, respectively. The proposed method also shows around threefold data efficiency improvements on the Potsdam dataset and domain generalization on the Enschede datasets. These results demonstrate the effectiveness of the proposed method in height map estimation from single-view remote sensing images. [ABSTRACT FROM AUTHOR]

Deep learning (DL) is currently the dominant approach to image classification and segmentation, but the performances of DL methods are remarkably influenced by the quantity and quality of the ground truth (GT) used for training. In this article, a DL method is presented to deal with the semantic segmentation of very-high-resolution (VHR) remote-sensing data in the case of scarce GT. The main idea is to combine a specific type of deep convolutional neural networks (CNNs), namely fully convolutional networks (FCNs), with probabilistic graphical models (PGMs). Our method takes advantage of the intrinsic multiscale behavior of FCNs to deal with multiscale data representations and to connect them to a hierarchical Markov model (e.g., making use of a quadtree). As a consequence, the spatial information present in the data is better exploited, allowing a reduced sensitivity to GT incompleteness to be obtained. The marginal posterior mode (MPM) criterion is used for inference in the proposed framework. To assess the capabilities of the proposed method, the experimental validation is conducted with the ISPRS 2D Semantic Labeling Challenge datasets on the cities of Vaihingen and Potsdam, with some modifications to simulate the spatially sparse GTs that are common in real remote-sensing applications. The results are quite significant, as the proposed approach exhibits a higher producer accuracy than the standard FCNs considered and especially mitigates the impact of scarce GTs on minority classes and small spatial details. [ABSTRACT FROM AUTHOR]

Currently, an increasing number of convolutional neural networks (CNNs) focus specifically on capturing contextual features (con. feat) to improve performance in semantic segmentation tasks. However, high-level con. feat are biased towards encoding features of large objects, disregard spatial details, and have a limited capacity to discriminate between easily confused classes (e.g., trees and grasses). As a result, we incorporate low-level features (low. feat) and class-specific discriminative features (dis. feat) to boost model performance further, with low. feat helping the model in recovering spatial information and dis. feat effectively reducing class confusion during segmentation. To this end, we propose a novel deep multi-feature learning framework for the semantic segmentation of VHR RSIs, dubbed MFNet. The proposed MFNet adopts a multi-feature learning mechanism to learn more complete features, including con. feat, low. feat, and dis. feat. More specifically, aside from a widely used context aggregation module for capturing con. feat, we additionally append two branches for learning low. feat and dis. feat. One focuses on learning low. feat at a shallow layer in the backbone network through local contrast processing, while the other groups con. feat and then optimizes each class individually to generate dis. feat with better inter-class discriminative capability. Extensive quantitative and qualitative evaluations demonstrate that the proposed MFNet outperforms most state-of-the-art models on the ISPRS Vaihingen and Potsdam datasets. In particular, thanks to the mechanism of multi-feature learning, our model achieves an overall accuracy score of 91.91% on the Potsdam test set with VGG16 as a backbone, performing favorably against advanced models with ResNet101. [ABSTRACT FROM AUTHOR]

Semantic segmentation of remote sensing imagery is a fundamental task in intelligent interpretation. Since deep convolutional neural networks (DCNNs) performed considerable insight in learning implicit representations from data, numerous works in recent years have transferred the DCNN-based model to remote sensing data analysis. However, the wide-range observation areas, complex and diverse objects and illumination and imaging angle influence the pixels easily confused, leading to undesirable results. Therefore, a remote sensing imagery semantic segmentation neural network, named HCANet, is proposed to generate representative and discriminative representations for dense predictions. HCANet hybridizes cross-level contextual and attentive representations to emphasize the distinguishability of learned features. First of all, a cross-level contextual representation module (CCRM) is devised to exploit and harness the superpixel contextual information. Moreover, a hybrid representation enhancement module (HREM) is designed to fuse cross-level contextual and self-attentive representations flexibly. Furthermore, the decoder incorporates DUpsampling operation to boost the efficiency losslessly. The extensive experiments are implemented on the Vaihingen and Potsdam benchmarks. In addition, the results indicate that HCANet achieves excellent performance on overall accuracy and mean intersection over union. In addition, the ablation study further verifies the superiority of CCRM. [ABSTRACT FROM AUTHOR]

In this work, we propose a new deep convolution neural network (DCNN) architecture for semantic segmentation of aerial imagery. Taking advantage of recent research, we use split-attention networks (ResNeSt) as the backbone for high-quality feature expression. Additionally, a disentangled nonlocal (DNL) block is integrated into our pipeline to express the inter-pixel long-distance dependence and highlight the edge pixels simultaneously. Moreover, the depth-wise separable convolution and atrous spatial pyramid pooling (ASPP) modules are combined to extract and fuse multiscale contextual features. Finally, an auxiliary edge detection task is designed to provide edge constraints for semantic segmentation. Evaluation of algorithms is conducted on two benchmarks provided by the International Society for Photogrammetry and Remote Sensing (ISPRS). Extensive experiments demonstrate the effectiveness of each module of our architecture. Precision evaluation based on the Potsdam benchmark shows that the proposed DCNN achieves competitive performance over the state-of-the-art methods. [ABSTRACT FROM AUTHOR]

Deep learning methods have been used to extract buildings from remote sensing images and have achieved state-of-the-art performance. Most previous work has emphasized the multi-scale fusion of features or the enhancement of more receptive fields to achieve global features rather than focusing on low-level details such as the edges. In this work, we propose a novel end-to-end edge-aware network, the EANet, and an edge-aware loss for getting accurate buildings from aerial images. Specifically, the architecture is composed of image segmentation networks and edge perception networks that, respectively, take charge of building prediction and edge investigation. The International Society for Photogrammetry and Remote Sensing (ISPRS) Potsdam segmentation benchmark and the Wuhan University (WHU) building benchmark were used to evaluate our approach, which, respectively, was found to achieve 90.19% and 93.33% intersection-over-union and top performance without using additional datasets, data augmentation, and post-processing. The EANet is effective in extracting buildings from aerial images, which shows that the quality of image segmentation can be improved by focusing on edge details. [ABSTRACT FROM AUTHOR]

Dense semantic labeling plays a pivotal role in high-resolution remote sensing image research. It provides pixel-level classification which is crucial in land cover mapping and urban planning. With the recent success of the convolutional neural network (CNN), accuracy has been greatly improved by previous works. However, most networks boost performance by involving too many parameters and computational overheads, which results in more inference time and hardware resources, while some attempts with light-weight networks do not achieve satisfactory results due to the insufficient feature extraction ability. In this work, we propose an efficient light-weight CNN based on dual-path architecture to address this issue. Our model utilizes three convolution layers as the spatial path to enhance the extraction of spatial information. Meanwhile, we develop the context path with the multi-fiber network (MFNet) followed by the pyramid pooling module (PPM) to obtain a sufficient receptive field. On top of these two paths, we adopt the channel attention block to refine the features from the context path and apply a feature fusion module to combine spatial information with context information. Moreover, a weighted cascade loss function is employed to enhance the learning procedure. With all these components, the performance can be significantly improved. Experiments on the Potsdam and Vaihingen datasets demonstrate that our network performs better than other light-weight networks, even some classic networks. Compared to the state-of-the-art U-Net, our model achieves higher accuracy on the two datasets with 2.5 times less network parameters and 22 times less computational floating point operations (FLOPs). [ABSTRACT FROM AUTHOR]

Semantic segmentation on high-resolution aerial images plays a significant role in many remote sensing applications. Although the Deep Convolutional Neural Network (DCNN) has shown great performance in this task, it still faces the following two challenges: intra-class heterogeneity and inter-class homogeneity. To overcome these two problems, a novel dual-path DCNN, which contains a spatial path and an edge path, is proposed for high-resolution aerial image segmentation. The spatial path, which combines the multi-level and global context features to encode the local and global information, is used to address the intra-class heterogeneity challenge. For inter-class homogeneity problem, a Holistically-nested Edge Detection (HED)-like edge path is employed to detect the semantic boundaries for the guidance of feature learning. Furthermore, we improve the computational efficiency of the network by employing the backbone of MobileNetV2. We enhance the performance of MobileNetV2 with two modifications: (1) replacing the standard convolution in the last four Bottleneck Residual Blocks (BRBs) with atrous convolution; and (2) removing the convolution stride of 2 in the first layer of BRBs 4 and 6. Experimental results on the ISPRS Vaihingen and Potsdam 2D labeling dataset show that the proposed DCNN achieved real-time inference speed on a single GPU card with better performance, compared with the state-of-the-art baselines. [ABSTRACT FROM AUTHOR]

In this study, a deep learning (DL)-based approach is proposed for the detection and reconstruction of buildings from a single aerial image. The pre-required knowledge to reconstruct the 3D shapes of buildings, including the height data as well as the linear elements of individual roofs, is derived from the RGB image using an optimized multi-scale convolutional–deconvolutional network (MSCDN). The proposed network is composed of two feature extraction levels to first predict the coarse features, and then automatically refine them. The predicted features include the normalized digital surface models (nDSMs) and linear elements of roofs in three classes of eave, ridge, and hip lines. Then, the prismatic models of buildings are generated by analyzing the eave lines. The parametric models of individual roofs are also reconstructed using the predicted ridge and hip lines. The experiments show that, even in the presence of noises in height values, the proposed method performs well on 3D reconstruction of buildings with different shapes and complexities. The average root mean square error (RMSE) and normalized median absolute deviation (NMAD) metrics are about 3.43 m and 1.13 m, respectively for the predicted nDSM. Moreover, the quality of the extracted linear elements is about 91.31% and 83.69% for the Potsdam and Zeebrugge test data, respectively. Unlike the state-of-the-art methods, the proposed approach does not need any additional or auxiliary data and employs a single image to reconstruct the 3D models of buildings with the competitive precision of about 1.2 m and 0.8 m for the horizontal and vertical RMSEs over the Potsdam data and about 3.9 m and 2.4 m over the Zeebrugge test data. [ABSTRACT FROM AUTHOR]