Search

In recent years, deep learning has found widespread application in SAR image object detection. However, when detecting multi-scale targets against complex backgrounds, these models often struggle to strike a balance between accuracy and speed. Furthermore, there is a continuous need to enhance the performance of current models. Hence, this paper proposes LRMSNet, a new multi-scale target detection model designed specifically for SAR images in complex backgrounds. Firstly, the paper introduces an attention module designed to enhance contextual information aggregation and capture global features, which is integrated into a backbone network with an expanded receptive field for improving SAR image feature extraction. Secondly, this paper develops an information aggregation module to effectively fuse different feature layers of the backbone network. Lastly, to better integrate feature information at various levels, this paper designs a multi-scale aggregation network. We validate the effectiveness of our method on three different SAR object detection datasets (MSAR-1.0, SSDD, and HRSID). Experimental results demonstrate that LRMSNet achieves outstanding performance with a mean average accuracy (mAP) of 95.2%, 98.9%, and 93.3% on the MSAR-1.0, SSDD, and HRSID datasets, respectively, with only 3.46 M parameters and 12.6 G floating-point operation cost (FLOPs). When compared with existing SAR object detection models on the MSAR-1.0 dataset, LRMSNet achieves state-of-the-art (SOTA) performance, showcasing its superiority in addressing SAR detection challenges in large-scale complex environments and across various object scales. [ABSTRACT FROM AUTHOR]

Underwater acoustic (UWA) channel prediction technology, as an important topic in UWA communication, has played an important role in UWA adaptive communication network and underwater target perception. Although many significant advancements have been achieved in underwater acoustic channel prediction over the years, a comprehensive summary and introduction is still lacking. As the first comprehensive overview of UWA channel prediction, this paper introduces past works and algorithm implementation methods of channel prediction from the perspective of linear, kernel-based, and deep learning approaches. Importantly, based on available at-sea experiment datasets, this paper compares the performance of current primary UWA channel prediction algorithms under a unified system framework, providing researchers with a comprehensive and objective understanding of UWA channel prediction. Finally, it discusses the directions and challenges for future research. The survey finds that linear prediction algorithms are the most widely applied, and deep learning, as the most advanced type of algorithm, has moved this field into a new stage. The experimental results show that the linear algorithms have the lowest computational complexity, and when the training samples are sufficient, deep learning algorithms have the best prediction performance. [ABSTRACT FROM AUTHOR]

Recently, deep learning techniques have been extensively used to detect ships in synthetic aperture radar (SAR) images. The majority of modern algorithms can achieve successful ship detection outcomes when working with multiple-scale ships on a large sea surface. However, there are still issues, such as missed detection and incorrect identification when performing multi-scale ship object detection operations in SAR images of complex scenes. To solve these problems, this paper proposes a complex scenes multi-scale ship detection model, according to YOLOv7, called CSD-YOLO. First, this paper suggests an SAS-FPN module that combines atrous spatial pyramid pooling and shuffle attention, allowing the model to focus on important information and ignore irrelevant information, reduce the feature loss of small ships, and simultaneously fuse the feature maps of ship targets on various SAR image scales, thereby improving detection accuracy and the model's capacity to detect objects at several scales. The model's optimization is then improved with the aid of the SIoU loss function. Finally, thorough tests on the HRSID and SSDD datasets are presented to support our methodology. CSD-YOLO achieves better detection performance than the baseline YOLOv7, with a 98.01% detection accuracy, a 96.18% recall, and a mean average precision (mAP) of 98.60% on SSDD. In addition, in comparative experiments with other deep learning-based methods, in terms of overall performance, CSD-YOLO still performs better. [ABSTRACT FROM AUTHOR]

Researchers have explored various methods to fully exploit the all-weather characteristics of Synthetic aperture radar (SAR) images to achieve high-precision, real-time, computationally efficient, and easily deployable ship target detection models. These methods include Constant False Alarm Rate (CFAR) algorithms and deep learning approaches such as RCNN, YOLO, and SSD, among others. While these methods outperform traditional algorithms in SAR ship detection, challenges still exist in handling the arbitrary ship distributions and small target features in SAR remote sensing images. Existing models are complex, with a large number of parameters, hindering effective deployment. This paper introduces a YOLOv7 oriented bounding box SAR ship detection model (YOLOv7oSAR). The model employs a rotation box detection mechanism, uses the KLD loss function to enhance accuracy, and introduces a Bi-former attention mechanism to improve small target detection. By redesigning the network's width and depth and incorporating a lightweight P-ELAN structure, the model effectively reduces its size and computational requirements. The proposed model achieves high-precision detection results on the public RSDD dataset (94.8% offshore, 66.6% nearshore), and its generalization ability is validated on a custom dataset (94.2% overall detection accuracy). [ABSTRACT FROM AUTHOR]

Due to the issues of remote sensing object detection algorithms based on deep learning, such as a high number of network parameters, large model size, and high computational requirements, it is challenging to deploy them on small mobile devices. This paper proposes an extremely lightweight remote sensing aircraft object detection network based on the improved YOLOv5n. This network combines Shufflenet v2 and YOLOv5n, significantly reducing the network size while ensuring high detection accuracy. It substitutes the original CIoU and convolution with EIoU and deformable convolution, optimizing for the small-scale characteristics of aircraft objects and further accelerating convergence and improving regression accuracy. Additionally, a coordinate attention (CA) mechanism is introduced at the end of the backbone to focus on orientation perception and positional information. We conducted a series of experiments, comparing our method with networks like GhostNet, PP-LCNet, MobileNetV3, and MobileNetV3s, and performed detailed ablation studies. The experimental results on the Mar20 public dataset indicate that, compared to the original YOLOv5n network, our lightweight network has only about one-fifth of its parameter count, with only a slight decrease of 2.7% in mAP@0.5. At the same time, compared with other lightweight networks of the same magnitude, our network achieves an effective balance between detection accuracy and resource consumption such as memory and computing power, providing a novel solution for the implementation and hardware deployment of lightweight remote sensing object detection networks. [ABSTRACT FROM AUTHOR]

Deep learning, which is a dominating technique in artificial intelligence, has completely changed image understanding over the past decade. As a consequence, the sea ice extraction (SIE) problem has reached a new era. We present a comprehensive review of four important aspects of SIE, including algorithms, datasets, applications and future trends. Our review focuses on research published from 2016 to the present, with a specific focus on deep-learning-based approaches in the last five years. We divided all related algorithms into three categories, including the conventional image classification approach, the machine learning-based approach and deep-learning-based methods. We reviewed the accessible ice datasets including SAR-based datasets, the optical-based datasets and others. The applications are presented in four aspects including climate research, navigation, geographic information systems (GIS) production and others. This paper also provides insightful observations and inspiring future research directions. [ABSTRACT FROM AUTHOR]

As one of the most important techniques for hyperspectral image dimensionality reduction, band selection has received considerable attention, whereas self-representation subspace clustering-based band selection algorithms have received quite a lot of attention with good effect. However, many of them lack the self-supervision of representations and ignore the multi-level spectral–spatial information of HSI and the connectivity of subspaces. To this end, this paper proposes a novel self-supervised multi-level representation learning fusion-based maximum entropy subspace clustering (MLRLFMESC) method for hyperspectral band selection. Firstly, to learn multi-level spectral–spatial information, self-representation subspace clustering is embedded between the encoder layers of the deep-stacked convolutional autoencoder and its corresponding decoder layers, respectively, as multiple fully connected layers to achieve multi-level representation learning (MLRL). A new auxiliary task is constructed for multi-level representation learning and multi-level self-supervised training to improve its capability of representation. Then, a fusion model is designed to fuse the multi-level spectral–spatial information to obtain a more distinctive coefficient matrix for self-expression, where the maximum entropy regularization (MER) method is employed to promote connectivity and the uniform dense distribution of band elements in each subspace. Finally, subspace clustering is conducted to obtain the final band subset. Experiments have been conducted on three hyperspectral datasets, and the corresponding results show that the proposed MLRLFMESC algorithm significantly outperforms several other band selection methods in classification performance. [ABSTRACT FROM AUTHOR]

The field of remote sensing information processing places significant research emphasis on object detection (OD) in high-spatial-resolution remote sensing images (HSRIs). The OD task in HSRIs poses additional challenges compared to conventional natural images. These challenges include variations in object scales, complex backgrounds, dense arrangement, and uncertain orientations. These factors contribute to the increased difficulty of OD in HSRIs as compared to conventional images. To tackle the aforementioned challenges, this paper introduces an innovative OD algorithm that builds upon enhancements made to the YOLOv5 framework. The incorporation of RepConv, Transformer Encoder, and BiFPN modules into the original YOLOv5 network leads to improved detection accuracy, particularly for objects of varying scales. The C3GAM module is designed by introducing the GAM attention mechanism to address the interference caused by complex background regions. To achieve precise localization of densely arranged objects, the SIoU loss function is integrated into YOLOv5. The circular smooth label method is used to detect objects with uncertain directions. The effectiveness of the suggested algorithm is confirmed through its application to two commonly utilized datasets, specifically HRSC2016 and UCAS-AOD. The average detection accuracies achieved on these datasets are 90.29% and 90.06% respectively, surpassing the performance of other compared OD algorithms for HSRIs. [ABSTRACT FROM AUTHOR]

In Mars exploration, hyper-spectrometry plays an important role due to its high spectral resolution. However, due to the technical difficulty and the data size, the spatial resolution or the coverage of hyperspectral data is often limited. This limitation can be alleviated by deep learning-based super-resolution (SR) reconstruction. But the spatial size and batch size of the input training data is limited due to the large number of spectral channels. To improve the efficiency of model training and SR reconstruction, a dataset based on CRISM hyperspectral data is created in this paper, and its redundancy is analyzed in both spectral and spatial spital dimensions. Compression algorithms based on data selection and PCA are used to reduce the size of the input training data. A network that can perform spatial SR and spectral enhancement is also proposed to make the network can be trained with the compressed data. With these compression algorithms and network, high-resolution data with 235 bands can be reconstructed from the low-resolution data with only 40 bands. Compared with the network trained on the original low-resolution data with 235 bands, the model training time and the SR reconstruction runtime can be reduced to 30% and 23% with practically no accuracy loss. The effectiveness of compression algorithms based on data selection also indicates that maybe not all the bands need to be transmitted from the Mars probes or be collected. Furthermore, it would, in principle, help improve the efficiency of satellite data transmission and simplify the design of the hyper-spectrometer. Additionally, a method for spatial dimension correlation evaluation is also proposed in this paper. The spatial compression shows that the proposed method can reflect the correlation of spatial texture between patches, and the model can be acceptably trained with only half of the original data. [ABSTRACT FROM AUTHOR]

Small target detection has been widely used in applications that are relevant to everyday life and have many real-time requirements, such as road patrols and security surveillance. Although object detection methods based on deep learning have achieved great success in recent years, they are not effective in small target detection. In order to solve the problem of low recognition rate caused by factors such as low resolution of UAV viewpoint images and little valid information, this paper proposes an improved algorithm based on the YOLOv5s model, called YOLOv5s-pp. First, to better suppress interference from complex backgrounds and negative samples in images, we add a CA attention module, which can better focus on task-specific important channels while weakening the influence of irrelevant channels. Secondly, we improve the forward propagation and generalisation of the network using the Meta-ACON activation function, which adaptively learns to adjust the degree of linearity or nonlinearity of the activation function based on the input data. Again, the SPD Conv module is incorporated into the network model to address the problems of reduced learning efficiency and loss of fine-grained information due to cross-layer convolution in the model. Finally, the detection head is improved by using smaller, smaller-target detection heads to reduce missed detections. We evaluated the algorithm on the VisDrone2019-DET and UAVDT datasets and compared it with other state-of-the-art algorithms. Compared to YOLOv5s, mAP@.5 improved by 7.4% and 6.5% on the VisDrone2019-DET and UAVDT datasets, respectively, and compared to YOLOv8s, mAP@.5 improved by 0.8% and 2.1%, respectively. For improving the performance of the UAV-side small target detection algorithm, it will help to enhance the reliability and safety of UAVs in critical missions such as military reconnaissance, road patrol and security surveillance. [ABSTRACT FROM AUTHOR]

The "low, slow, and small" target (LSST) poses a significant threat to the military ground unit. It is hard to defend against due to its invisibility to numerous detecting devices. With the onboard deep learning-based object detection methods, the intelligent LSST (ILSST) can find and detect the ground unit autonomously in a denied environment. This paper proposes an adversarial patch-based defending method to blind the ILSST by attacking its onboard object detection network. First, an adversarial influence score was established to indicate the influence of the adversarial noise on the objects. Then, based on this score, we used the least squares algorithm and Bisectional search methods to search the patch's optimal coordinates and size. Using the optimal coordinates and size, an adaptive patch-generating network was constructed to automatically generate patches on ground units and hide the ground units from the deep learning-based object detection network. To evaluate the efficiency of our algorithm, a new LSST view dataset was collected, and extensive attacking experiments are carried out on this dataset. The results demonstrate that our algorithm can effectively attack the object detection networks, is better than state-of-the-art adversarial patch-generating algorithms in hiding the ground units from the object detection networks, and has high transferability among the object detection networks. [ABSTRACT FROM AUTHOR]

This paper aims to develop a multi-rotor-based visual tracker for a specified moving object. Visual object-tracking algorithms for multi-rotors are challenging due to multiple issues such as occlusion, quick camera motion, and out-of-view scenarios. Hence, algorithmic changes are required for dealing with images or video sequences obtained by multi-rotors. Therefore, we propose two approaches: a generic object tracker and a class-specific tracker. Both tracking settings require the object bounding box to be selected in the first frame. As part of the later steps, the object tracker uses the updated template set and the calibrated RGBD sensor data as inputs to track the target object using a Siamese network and a machine-learning model for depth estimation. The class-specific tracker is quite similar to the generic object tracker but has an additional auxiliary object classifier. The experimental study and validation were carried out in a robot simulation environment. The simulation environment was designed to serve multiple case scenarios using Gazebo. According to the experiment results, the class-specific object tracker performed better than the generic object tracker in terms of stability and accuracy. Experiments show that the proposed generic tracker achieves promising results on three challenging datasets. Our tracker runs at approximately 36 fps on GPU. [ABSTRACT FROM AUTHOR]

Most background suppression algorithms are weakly robust due to the complexity and fluctuation of the star image's background. In this paper, a background suppression algorithm for stray lights in star images is proposed, which is named BSC-Net (Background Suppression Convolutional Network) and consist of two parts: "Background Suppression Part" and "Foreground Retention Part". The former part achieves background suppression by extracting features from various receptive fields, while the latter part achieves foreground retention by merging multi-scale features. Through this two-part design, BSC-Net can compensate for blurring and distortion of the foreground caused by background suppression, which is not achievable in other methods. At the same time, a blended loss function of smooth_L1&Structure Similarity Index Measure (SSIM) is introduced to hasten the network convergence and avoid image distortion. Based on the BSC-Net and the loss function, a dataset consisting of real images will be used for training and testing. Finally, experiments show that BSC-Net achieves the best results and the largest Signal-to-Noise Ratio (SNR) improvement in different backgrounds, which is fast, practical and efficient, and can tackle the shortcomings of existing methods. [ABSTRACT FROM AUTHOR]

At present, deep learning has been widely used in SAR ship target detection, but the accurate and real-time detection of multi-scale targets still faces tough challenges. CNN-based SAR ship detectors are challenged to meet real-time requirements because of a large number of parameters. In this paper, we propose a lightweight, single-stage SAR ship target detection model called YOLO-based lightweight multi-scale ship detector (LMSD-YOLO), with better multi-scale adaptation capabilities. The proposed LMSD-YOLO consists of depthwise separable convolution, batch normalization and activate or not (ACON) activation function (DBA) module, Mobilenet with stem block (S-Mobilenet) backbone module, depthwise adaptively spatial feature fusion (DSASFF) neck module and SCYLLA-IoU (SIoU) loss function. Firstly, the DBA module is proposed as a general lightweight convolution unit to construct the whole lightweight model. Secondly, the improved S-Mobilenet module is designed as the backbone feature extraction network to enhance feature extraction ability without adding additional calculations. Then, the DSASFF module is proposed to achieve adaptive fusion of multi-scale features with fewer parameters. Finally, the SIoU is used as the loss function to accelerate model convergence and improve detection accuracy. The effectiveness of the LMSD-YOLO is validated on the SSDD, HRSID and GFSDD datasets, respectively, and the experimental results show that our proposed model has a smaller model volume and higher detection accuracy, and can accurately detect multi-scale targets in more complex scenes. The model volume of LMSD-YOLO is only 7.6MB (52.77% of model size of YOLOv5s), the detection speed on the NVIDIA AGX Xavier development board reached 68.3 FPS (32.7 FPS higher than YOLOv5s detector), indicating that the LMSD-YOLO can be easily deployed to the mobile platform for real-time application. [ABSTRACT FROM AUTHOR]

We study the problem of indoor positioning, which is a fundamental service in managing and analyzing objects in indoor environments. Unpredictable signal interference sources increase the degeneration of the accuracy and robustness of existing solutions. Deep learning approaches have recently been widely studied to overcome these challenges and attain better performance. In this paper, we aim to develop efficient algorithms, such as the dual-encoder-condensed convolution (DECC) method, which can achieve high-precision positioning for indoor services. In particular, firstly, we develop a convolutional module to add the original channel state information to the location information. Secondly, to explore channel differences between different antennas, we adopt a dual-encoder stacking mechanism for parallel calculation. Thirdly, we develop two different convolution kernels to speed up convergence. Performance studies on the indoor scenario and the urban canyon scenario datasets demonstrate the efficiency and effectiveness of our new approach. [ABSTRACT FROM AUTHOR]

The Loess Plateau is an ecologically fragile area in China; furthermore, loess landslides are typical forms of geological disasters, which severely limit the sustainable development of the local societies and the economy. Studying the automatic detection of landslides can facilitate disaster prevention and mitigation in the Loess Plateau, and help realize the climate action goal (SDG 13) of the United Nations Sustainable Development Goals (SDGs). This paper takes typical loess areas in China as the research object, and establishes a historical loess landslide sample database based on Google Earth (GEE) image data, with a total of 1451 loess landslides. The automatic detection of loess landslides is implemented by improving the You Only Look Once X (YOLOX) algorithm. The results show that the average accuracy of landslide detection in this method is 95.43%, and the accuracy rate is 96.32%, which effectively combines the earth's big data to realize the automatic detection of loess landslides. The research results provide technical support for the promotion of disaster prevention and mitigation in China's loess regions, the realization of sustainable development goals, and the improvement of natural disaster prevention–resistance–reduction systems. [ABSTRACT FROM AUTHOR]

Direction-of-arrival (DOA) estimation is still a pivotal research direction in array signal processing. Traditional algorithms based on the signal subspace and compressed sensing theory usually suffer from off-grid and computational complexity. Deep-learning-based methods usually face difficulty in obtaining labeled datasets. With the development of array technology, sparse sensor arrays can effectively reduce the number of sensors, which in turn reduces the complexity of the hardware. Therefore, effective DOA estimation algorithms for sparse sensor arrays need to be further investigated. An unsupervised deep learning method is proposed here to address the above issues. A training model was built based on the residual network structure. The DOA estimation was implemented using Vandermonde decomposition. Finally, the experimental findings confirmed the efficacy of the proposed algorithms presented in this article. [ABSTRACT FROM AUTHOR]

As an active microwave device, synthetic aperture radar (SAR) uses the backscatter of objects for imaging. SAR image ship targets are characterized by unclear contour information, a complex background and strong scattering. Existing deep learning detection algorithms derived from anchor-based methods mostly rely on expert experience to set a series of hyperparameters, and it is difficult to characterize the unique characteristics of SAR image ship targets, which greatly limits detection accuracy and speed. Therefore, this paper proposes a new lightweight position-enhanced anchor-free SAR ship detection algorithm called LPEDet. First, to resolve unclear SAR target contours and multiscale performance problems, we used YOLOX as the benchmark framework and redesigned the lightweight multiscale backbone, called NLCNet, which balances detection speed and accuracy. Second, for the strong scattering characteristics of the SAR target, we designed a new position-enhanced attention strategy, which suppresses background clutter by adding position information to the channel attention that highlights the target information to more accurately identify and locate the target. The experimental results for two large-scale SAR target detection datasets, SSDD and HRSID, show that our method achieves a higher detection accuracy and a faster detection speed than state-of-the-art SAR target detection methods. [ABSTRACT FROM AUTHOR]

In the process of using multi-channel ground-penetrating radar (GPR) for underground cavity exploration, the acquired 3D data include reflection data from underground cavities or various underground objects (structures). Reflection data from unspecified structures can interfere with the identification process of underground cavities. This study aims to identify underground cavities using a C-GAN model with an applied ResBlock technique. This deep learning model demonstrates excellent performance in the image domain and can automatically classify the presence of cavities by analyzing 3D GPR data, including reflection waveforms (A-scan), cross-sectional views (B-scan), and plan views (C-scan) measured from the ground under roads. To maximize the performance of the C-GAN model, a background filtering algorithm (BFA) was developed and applied to enhance the visibility and clarity of underground cavities. To verify the performance of the developed BFA, 3D data collected from roads in Seoul, Republic of Korea, using 3D GPR equipment were transformed, and the C-GAN model was applied. As a result, it was confirmed that the recall, an indicator of cavity prediction, improved by approximately 1.15 times compared to when the BFA was not applied. This signifies the verification of the effectiveness of the BFA. This study developed a special algorithm to distinguish underground cavities. This means that in the future, not only the advancement of separate equipment and systems but also the development of specific algorithms can contribute to the cavity exploration process. [ABSTRACT FROM AUTHOR]

Autonomous vehicle technology is advancing, with 3D object detection based on point clouds being crucial. However, point clouds' irregularity, sparsity, and large data volume, coupled with irrelevant background points, hinder detection accuracy. We propose a two-stage multi-scale 3D object detection network. Firstly, considering that a large number of useless background points are usually generated by the ground during detection, we propose a new ground filtering algorithm to increase the proportion of foreground points and enhance the accuracy and efficiency of the two-stage detection. Secondly, given that different types of targets to be detected vary in size, and the use of a single-scale voxelization may result in excessive loss of detailed information, the voxels of different scales are introduced to extract relevant features of objects of different scales in the point clouds and integrate them into the second-stage detection. Lastly, a multi-scale feature fusion module is proposed, which simultaneously enhances and integrates features extracted from voxels of different scales. This module fully utilizes the valuable information present in the point cloud across various scales, ultimately leading to more precise 3D object detection. The experiment is conducted on the KITTI dataset and the nuScenes dataset. Compared with our baseline, "Pedestrian" detection improved by 3.37–2.72% and "Cyclist" detection by 3.79–1.32% across difficulty levels on KITTI, and was boosted by 2.4% in NDS and 3.6% in mAP on nuScenes. [ABSTRACT FROM AUTHOR]

In recent years, deep learning techniques for pansharpening multiresolution images have gained increasing interest. Due to the lack of ground truth data, most deep learning solutions rely on synthetic reduced-resolution data for supervised training. This approach has limitations due to the statistical mismatch between real full-resolution and synthetic reduced-resolution data, which affects the models' generalization capacity. Consequently, there has been a shift towards unsupervised learning frameworks for pansharpening deep learning-based techniques. Unsupervised schemes require defining sophisticated loss functions with at least two components: one for spectral quality, ensuring consistency between the pansharpened image and the input multispectral component, and another for spatial quality, ensuring consistency between the output and the panchromatic input. Despite promising results, there has been limited investigation into the interaction and balance of these loss terms to ensure stability and accuracy. This work explores how unsupervised spatial and spectral consistency losses can be reliably combined preserving the outocome quality. By examining these interactions, we propose a general rule for balancing the two loss components to enhance the stability and performance of unsupervised pansharpening models. Experiments on three state-of-the-art algorithms using WorldView-3 images demonstrate that methods trained with the proposed framework achieve good performance in terms of visual quality and numerical indexes. [ABSTRACT FROM AUTHOR]

With success of Deep Belief Networks (DBNs) in computer vision, DBN has attracted great attention in hyperspectral classification. Many deep learning based algorithms have been focused on deep feature extraction for classification improvement. Multi-features, such as texture feature, are widely utilized in classification process to enhance classification accuracy greatly. In this paper, a novel hyperspectral classification framework based on an optimal DBN and a novel texture feature enhancement (TFE) is proposed. Through band grouping, sample band selection and guided filtering, the texture features of hyperspectral data are improved. After TFE, the optimal DBN is employed on the hyperspectral reconstructed data for feature extraction and classification. Experimental results demonstrate that the proposed classification framework outperforms some state-of-the-art classification algorithms, and it can achieve outstanding hyperspectral classification performance. Furthermore, our proposed TFE method can play a significant role in improving classification accuracy. [ABSTRACT FROM AUTHOR]

Detecting oriented small objects is a critical task in remote sensing, but the development of high-performance deep learning-based detectors is hindered by the need for large-scale and well-annotated datasets. The high cost of creating these datasets, due to the dense and numerous distribution of small objects, significantly limits the application and development of such detectors. To address this problem, we propose a single-point-based annotation approach (SPA) based on the graph cut method. In this framework, user annotations act as the origin of positive sample points, and a similarity matrix, computed from feature maps extracted by deep learning networks, facilitates an intuitive and efficient annotation process for building graph elements. Utilizing the Maximum Flow algorithm, SPA derives positive sample regions from these points and generates oriented bounding boxes (OBBOXs). Experimental results demonstrate the effectiveness of SPA, with at least a 50% improvement in annotation efficiency. Furthermore, the intersection-over-union (IoU) metric of our OBBOX is 3.6% higher than existing methods such as the "Segment Anything Model". When applied in training, the model annotated with SPA shows a 4.7% higher mean average precision (mAP) compared to models using traditional annotation methods. These results confirm the technical advantages and practical impact of SPA in advancing small object detection in remote sensing. [ABSTRACT FROM AUTHOR]

Automatic driving technology is becoming one of the main areas of development for future intelligent transportation systems. The high-precision map, which is an important supplement of the on-board sensors during shielding or limited observation distance, provides a priori information for high-precision positioning and path planning in automatic driving. The position and semantic information of the road markings, such as absolute coordinates of the solid lines and dashed lines, are the basic components of the high-precision map. In this paper, we study the automatic extraction and vectorization of road markings. Firstly, scan lines are extracted from the vehicle-borne laser point cloud data, and the pavement is extracted from scan lines according to the geometric mutation at the road boundary. On this basis, the pavement point clouds are transformed into raster images with a certain resolution by using the method of inverse distance weighted interpolation. An adaptive threshold segmentation algorithm is used to convert raster images into binary images. Followed by the adaptive threshold segmentation is the Euclidean clustering method, which is used to extract road markings point clouds from the binary image. Solid lines are detected by feature attribute filtering. All of the solid lines and guidelines in the sample data are correctly identified. The deep learning network framework PointNet++ is used for semantic recognition of the remaining road markings, including dashed lines, guidelines and arrows. Finally, the vectorization of the identified solid lines and dashed lines is carried out based on a line segmentation self-growth algorithm. The vectorization of the identified guidelines is carried out according to an alpha shape algorithm. Point cloud data from four experimental areas are used for road marking extraction and identification. The F-scores of the identification of dashed lines, guidelines, straight arrows and right turn arrows are 0.97, 0.66, 0.84 and 1, respectively. [ABSTRACT FROM AUTHOR]

Landsat cloud and cloud shadow detection has a long heritage based on the application of empirical spectral tests to single image pixels, including the Landsat product Fmask algorithm, which uses spectral tests applied to optical and thermal bands to detect clouds and uses the sun-sensor-cloud geometry to detect shadows. Since the Fmask was developed, convolutional neural network (CNN) algorithms, and in particular U-Net algorithms (a type of CNN with a U-shaped network structure), have been developed and are applied to pixels in square patches to take advantage of both spatial and spectral information. The purpose of this study was to develop and assess a new U-Net algorithm that classifies Landsat 8/9 Operational Land Imager (OLI) pixels with higher accuracy than the Fmask algorithm. The algorithm, termed the Learning Attention Network Algorithm (LANA), is a form of U-Net but with an additional attention mechanism (a type of network structure) that, unlike conventional U-Net, uses more spatial pixel information across each image patch. The LANA was trained using 16,861 512 × 512 30 m pixel annotated Landsat 8 OLI patches extracted from 27 images and 69 image subsets that are publicly available and have been used by others for cloud mask algorithm development and assessment. The annotated data were manually refined to improve the annotation and were supplemented with another four annotated images selected to include clear, completely cloudy, and developed land images. The LANA classifies image pixels as either clear, thin cloud, cloud, or cloud shadow. To evaluate the classification accuracy, five annotated Landsat 8 OLI images (composed of >205 million 30 m pixels) were classified, and the results compared with the Fmask and a publicly available U-Net model (U-Net Wieland). The LANA had a 78% overall classification accuracy considering cloud, thin cloud, cloud shadow, and clear classes. As the LANA, Fmask, and U-Net Wieland algorithms have different class legends, their classification results were harmonized to the same three common classes: cloud, cloud shadow, and clear. Considering these three classes, the LANA had the highest (89%) overall accuracy, followed by Fmask (86%), and then U-Net Wieland (85%). The LANA had the highest F1-scores for cloud (0.92), cloud shadow (0.57), and clear (0.89), and the other two algorithms had lower F1-scores, particularly for cloud (Fmask 0.90, U-Net Wieland 0.88) and cloud shadow (Fmask 0.45, U-Net Wieland 0.52). In addition, a time-series evaluation was undertaken to examine the prevalence of undetected clouds and cloud shadows (i.e., omission errors). The band-specific temporal smoothness index (TSIλ) was applied to a year of Landsat 8 OLI surface reflectance observations after discarding pixel observations labelled as cloud or cloud shadow. This was undertaken independently at each gridded pixel location in four 5000 × 5000 30 m pixel Landsat analysis-ready data (ARD) tiles. The TSIλ results broadly reflected the classification accuracy results and indicated that the LANA had the smallest cloud and cloud shadow omission errors, whereas the Fmask had the greatest cloud omission error and the second greatest cloud shadow omission error. Detailed visual examination, true color image examples and classification results are included and confirm these findings. The TSIλ results also highlight the need for algorithm developers to undertake product quality assessment in addition to accuracy assessment. The LANA model, training and evaluation data, and application codes are publicly available for other researchers. [ABSTRACT FROM AUTHOR]

Three-dimensional object detection is a pivotal research topic in computer vision, aiming to identify and locate objects in three-dimensional space. It has wide applications in various fields such as geoscience, autonomous driving, and drone navigation. The rapid development of deep learning techniques has led to significant advancements in 3D object detection. However, with the increasing complexity of applications, 3D object detection faces a series of challenges such as data imbalance and the effectiveness of network models. Specifically, in an experiment, our investigation revealed a notable discrepancy in the LiDAR reflection intensity within a point cloud scene, with stronger intensities observed in proximity and weaker intensities observed at a distance. Furthermore, we have also noted a substantial disparity in the number of foreground points compared to the number of background points. Especially in 3D object detection, the foreground point is more important than the background point, but it is usually downsampled without discrimination in the subsequent processing. With the objective of tackling these challenges, we work from both data and network perspectives, designing a feature alignment filtering algorithm and a two-stage 3D object detection network. Firstly, in order to achieve feature alignment, we introduce a correction equation to decouple the relationship between distance and intensity and eliminate the attenuation effect of intensity caused by distance. Then, a background point filtering algorithm is designed by using the aligned data to alleviate the problem of data imbalance. At the same time, we take into consideration the fact that the accuracy of semantic segmentation plays a crucial role in 3D object detection. Therefore, we propose a two-stage deep learning network that integrates spatial and spectral information, in which a feature fusion branch is designed and embedded in the semantic segmentation backbone. Through a series of experiments on the KITTI dataset, it is proven that the proposed method achieves the following average precision (AP_R40) values for easy, moderate, and hard difficulties, respectively: car (Iou 0.7)—89.23%, 80.14%, and 77.89%; pedestrian (Iou 0.5)—52.32%, 45.47%, and 38.78%; and cyclist (Iou 0.5)—76.41%, 61.92%, and 56.39%. By emphasizing both data quality optimization and efficient network architecture, the performance of the proposed method is made comparable to other state-of-the-art methods. [ABSTRACT FROM AUTHOR]

Band selection (BS) is an efficacious approach to reduce hyperspectral information redundancy while preserving the physical meaning of hyperspectral images (HSIs). Recently, deep learning-based BS methods have received widespread interest due to their ability to model the nonlinear relationship between bands, with existing methods typically relying on generative algorithms. However, the process of generating images with pixel-level detail required by generative algorithm-based BS methods is computationally expensive. To alleviate this issue, we propose a contrastive learning-based unsupervised BS architecture, termed ContrastBS, in this article. With the help of contrastive learning, the proposed architecture avoids the costly generation step in pixel space by learning to distinguish data at the abstract semantic level of the feature space. Specifically, ContrastBS combines an attention mechanism with contrastive learning to extract the importance of each band. Furthermore, we design a novel loss function, which is able to constrain the symmetric loss while ensuring attention to the most valuable bands, for the contrastive learning-based BS network. Experimental results indicate that ContrastBS has excellent classification performance and competitive time cost compared to the comparison methods. [ABSTRACT FROM AUTHOR]

Hyperspectral image (HSI) change detection plays an important role in remote sensing applications, and considerable research has been done focused on improving change detection performance. However, the high dimension of hyperspectral data makes it hard to extract discriminative features for hyperspectral processing tasks. Though deep convolutional neural networks (CNN) have superior capability in high-level semantic feature learning, it is difficult to employ CNN for change detection tasks. As a ground truth map is usually used for the evaluation of change detection algorithms, it cannot be directly used for supervised learning. In order to better extract discriminative CNN features, a novel noise modeling-based unsupervised fully convolutional network (FCN) framework is presented for HSI change detection in this paper. Specifically, the proposed method utilizes the change detection maps of existing unsupervised change detection methods to train the deep CNN, and then removes the noise during the end-to-end training process. The main contributions of this paper are threefold: (1) A new end-to-end FCN-based deep network architecture for HSI change detection is presented with powerful learning features; (2) An unsupervised noise modeling method is introduced for the robust training of the proposed deep network; (3) Experimental results on three datasets confirm the effectiveness of the proposed method. [ABSTRACT FROM AUTHOR]

Semantic segmentation is a fundamental task in remote sensing image analysis (RSIA). Fully convolutional networks (FCNs) have achieved state-of-the-art performance in the task of semantic segmentation of natural scene images. However, due to distinctive differences between natural scene images and remotely-sensed (RS) images, FCN-based semantic segmentation methods from the field of computer vision cannot achieve promising performances on RS images without modifications. In previous work, we proposed an RS image semantic segmentation framework SDFCNv1, combined with a majority voting postprocessing method. Nevertheless, it still has some drawbacks, such as small receptive field and large number of parameters. In this paper, we propose an improved semantic segmentation framework SDFCNv2 based on SDFCNv1, to conduct optimal semantic segmentation on RS images. We first construct a novel FCN model with hybrid basic convolutional (HBC) blocks and spatial-channel-fusion squeeze-and-excitation (SCFSE) modules, which occupies a larger receptive field and fewer network model parameters. We also put forward a data augmentation method based on spectral-specific stochastic-gamma-transform-based (SSSGT-based) during the model training process to improve generalizability of our model. Besides, we design a mask-weighted voting decision fusion postprocessing algorithm for image segmentation on overlarge RS images. We conducted several comparative experiments on two public datasets and a real surveying and mapping dataset. Extensive experimental results demonstrate that compared with the SDFCNv1 framework, our SDFCNv2 framework can increase the mIoU metric by up to 5.22% while only using about half of parameters. [ABSTRACT FROM AUTHOR]

This paper presents an algorithm for large-scale automatic detection of burial mounds, one of the most common types of archaeological sites globally, using LiDAR and multispectral satellite data. Although previous attempts were able to detect a good proportion of the known mounds in a given area, they still presented high numbers of false positives and low precision values. Our proposed approach combines random forest for soil classification using multitemporal multispectral Sentinel-2 data and a deep learning model using YOLOv3 on LiDAR data previously pre-processed using a multi–scale relief model. The resulting algorithm significantly improves previous attempts with a detection rate of 89.5%, an average precision of 66.75%, a recall value of 0.64 and a precision of 0.97, which allowed, with a small set of training data, the detection of 10,527 burial mounds over an area of near 30,000 km2, the largest in which such an approach has ever been applied. The open code and platforms employed to develop the algorithm allow this method to be applied anywhere LiDAR data or high-resolution digital terrain models are available. [ABSTRACT FROM AUTHOR]

Detecting animals to estimate abundance can be difficult, particularly when the habitat is dense or the target animals are fossorial. The recent surge in the use of thermal imagers in ecology and their use in animal detections can increase the accuracy of population estimates and improve the subsequent implementation of management programs. However, the use of thermal imagers results in many hours of captured flight videos which require manual review for confirmation of species detection and identification. Therefore, the perceived cost and efficiency trade-off often restricts the use of these systems. Additionally, for many off-the-shelf systems, the exported imagery can be quite low resolution (<9 Hz), increasing the difficulty of using automated detections algorithms to streamline the review process. This paper presents an animal species detection system that utilises the cost-effectiveness of these lower resolution thermal imagers while harnessing the power of transfer learning and an enhanced small object detection algorithm. We have proposed a distant object detection algorithm named Distant-YOLO (D-YOLO) that utilises YOLO (You Only Look Once) and improves its training and structure for the automated detection of target objects in thermal imagery. We trained our system on thermal imaging data of rabbits, their active warrens, feral pigs, and kangaroos collected by thermal imaging researchers in New South Wales and Western Australia. This work will enhance the visual analysis of animal species while performing well on low, medium and high-resolution thermal imagery. [ABSTRACT FROM AUTHOR]

Earth observation satellite task scheduling research plays a key role in space-based remote sensing services. An effective task scheduling strategy can maximize the utilization of satellite resources and obtain larger objective observation profits. In this paper, inspired by the success of deep reinforcement learning in optimization domains, the deep deterministic policy gradient algorithm is adopted to solve a time-continuous satellite task scheduling problem. Moreover, an improved graph-based minimum clique partition algorithm is proposed for preprocessing in the task clustering phase by considering the maximum task priority and the minimum observation slewing angle under constraint conditions. Experimental simulation results demonstrate that the deep reinforcement learning-based task scheduling method is feasible and performs much better than traditional metaheuristic optimization algorithms, especially in large-scale problems. [ABSTRACT FROM AUTHOR]

Deep learning-based algorithms for single object tracking (SOT) have shown impressive performance but remain susceptible to adversarial patch attacks. However, existing adversarial patch generation methods primarily focus on generating patches within the search region, neglecting the incorporation of template information, which limits their effectiveness in carrying out successful attacks. There is also a lack of evaluation metrics to assess the patch's adversarial abilities. In this study, we propose a bilateral adversarial patch-generating network to address these limitations and advance the field of adversarial patch generation for SOT networks. Our network leverages a Focus structure that effectively integrates both template and search region information, generating separate adversarial patches for each branch. We also introduce the DeFocus structure to solve the size discrepancy between the template and search region of the tracking network. To effectively mislead the tracking network, we have designed adversarial object loss and adversarial regression loss functions tailored to the network's output. Moreover, we propose a comprehensive evaluation metric that measures the patch's adversarial ability by establishing a relationship between the relative patch size and attack performance. As UAV view data often constitute small objects requiring smaller patches, we evaluate our approach on the UAV123 and UAVDT datasets. Our evaluation encompasses not only the overall attack performance but also the effectiveness of our strategy and the transferability of the attacks. Experimental results demonstrate that our algorithm generates patches with higher attack efficiency compared to existing methods. [ABSTRACT FROM AUTHOR]

Feature matching is a core step in multi-source remote sensing image registration approaches based on feature. However, for existing methods, whether traditional classical SIFT algorithm or deep learning-based methods, they essentially rely on generating descriptors from local regions of feature points, which can lead to low matching success rates due to various challenges, including gray-scale changes, content changes, local similarity, and occlusions between images. Inspired by the human approach of finding rough corresponding regions globally and then carefully comparing local regions, and the excellent global attention property of transformers, the proposed feature matching network adopts a coarse-to-fine matching strategy that utilizes both global and local information between images to predict corresponding feature points. Importantly, the network has great flexibility of matching corresponding points for any feature points and can be effectively trained without strong supervised signals of corresponding feature points and only require the true geometric transformation between images. The qualitative experiment illustrate the effectiveness of the proposed network by matching feature points extracted by SIFT or sampled uniformly. In the quantitative experiments, we used feature points extracted by SIFT, SuperPoint, and LoFTR as the keypoints to be matched. We then calculated the mean match success ratio (MSR) and mean reprojection error (MRE) of each method at different thresholds in the test dataset. Additionally, boxplot graphs were plotted to visualize the distributions. By comparing the MSR and MRE values as well as their distributions with other methods, we can conclude that the proposed method consistently outperforms the comparison methods in terms of MSR at different thresholds. Moreover, the MSR of the proposed method remains within a reasonable range compared to the MRE of other methods. [ABSTRACT FROM AUTHOR]

Cracks are fractures or breaks that occur in materials such as concrete, metals, rocks, and other solids. Various methods are used to detect and monitor cracks; among many of them, image-based methodologies allow fast identification of the distress and easy quantification of the percentage of cracks in the scene. Two main categories can be identified: classical and deep learning approaches. In the last decade, the tendency has moved towards the use of the latter. Even though they have proven their outstanding predicting performance, they suffer some drawbacks: a "black-box" nature leaves the user blind and without the possibility of modifying any parameters, a huge amount of labeled data is generally needed, a process that requires expert judgment is always required, and, finally, they tend to be time-consuming. Accordingly, the present study details the methodology for a new algorithm for crack segmentation based on the theory of minimal path selection combined with a region-based approach obtained through the segmentation of texture features extracted using Gabor filters. A pre-processing step is described, enabling the equalization of brightness and shadows, which results in better detection of local minima. These local minimal are constrained by a minimum distance between adjacent points, enabling a better coverage of the cracks. Afterward, a region-based segmentation technique is introduced to determine two areas that are used to determine threshold values used for rejection. This step is critical to generalize the algorithm to images presenting close-up scenes or wide cracks. Finally, a geometrical thresholding step is presented, allowing the exclusion of rounded areas and small isolated cracks. The results showed a very competitive F1-score (0.839), close to state-of-the-art values achieved with deep learning techniques. The main advantage of this approach is the transparency of the workflow, contrary to what happens with deep learning frameworks. In the proposed approach, no prior information is required; however, the statistical parameters may have to be adjusted to the particular case and requirements of the situation. The proposed algorithm results in a useful tool for researchers and practitioners needing to validate their results against some reference or needing labeled data for their models. Moreover, the current study could establish the grounds to standardize the procedure for crack segmentation with a lower human bias and faster results. The direct application of the methodology to images obtained with any low-cost sensor makes the proposed algorithm an operational support tool for authorities needing crack detection systems in order to monitor and evaluate the current state of the infrastructures, such as roads, tunnels, or bridges. [ABSTRACT FROM AUTHOR]

Accurate spatial information of agricultural fields is important for providing actionable information to farmers, managers, and policymakers. On the other hand, the automated detection of field boundaries is a challenging task due to their small size, irregular shape and the use of mixed-cropping systems making field boundaries vaguely defined. In this paper, we propose a strategy for field boundary detection based on the fully convolutional network architecture called ResU-Net. The benefits of this model are two-fold: first, residual units ease training of deep networks. Second, rich skip connections within the network could facilitate information propagation, allowing us to design networks with fewer parameters but better performance in comparison with the traditional U-Net model. An extensive experimental analysis is performed over the whole of Denmark using Sentinel-2 images and comparing several U-Net and ResU-Net field boundary detection algorithms. The presented results show that the ResU-Net model has a better performance with an average F1 score of 0.90 and average Jaccard coefficient of 0.80 in comparison to the U-Net model with an average F1 score of 0.88 and an average Jaccard coefficient of 0.77. [ABSTRACT FROM AUTHOR]

In this paper we introduce a novel machine learning-based fully automatic approach for the semantic analysis and documentation of masonry wall images, performing in parallel automatic detection and virtual completion of occluded or damaged wall regions, and brick segmentation leading to an accurate model of the wall structure. For this purpose, we propose a four-stage algorithm which comprises three interacting deep neural networks and a watershed transform-based brick outline extraction step. At the beginning, a U-Net-based sub-network performs initial wall segmentation into brick, mortar and occluded regions, which is followed by a two-stage adversarial inpainting model. The first adversarial network predicts the schematic mortar-brick pattern of the occluded areas based on the observed wall structure, providing in itself valuable structural information for archeological and architectural applications. The second adversarial network predicts the pixels' color values yielding a realistic visual experience for the observer. Finally, using the neural network outputs as markers in a watershed-based segmentation process, we generate the accurate contours of the individual bricks, both in the originally visible and in the artificially inpainted wall regions. Note that while the first three stages implement a sequential pipeline, they interact through dependencies of their loss functions admitting the consideration of hidden feature dependencies between the different network components. For training and testing the network a new dataset has been created, and an extensive qualitative and quantitative evaluation versus the state-of-the-art is given. The experiments confirmed that the proposed method outperforms the reference techniques both in terms of wall structure estimation and regarding the visual quality of the inpainting step, moreover it can be robustly used for various different masonry wall types. [ABSTRACT FROM AUTHOR]

Meteorological satellites are usually operated at high temporal resolutions, but the spatial resolutions are too poor to identify ground content. Super-resolution is an economic way to enhance spatial details, but the feasibility is not validated for meteorological images due to the absence of benchmarking data. In this work, we propose the FY4ASRgray and FY4ASRcolor datasets to assess super-resolution algorithms on meteorological applications. The features of cloud sensitivity and temporal continuity are linked to the proposed datasets. To test the usability of the new datasets, five state-of-the-art super-resolution algorithms are gathered for contest. Shift learning is used to shorten the training time and improve the parameters. Methods are modified to deal with the 16-bit challenge. The reconstruction results are demonstrated and evaluated regarding the radiometric, structural, and spectral loss, which gives the baseline performance for detail enhancement of the FY4A satellite images. Additional experiments are made on FY4ASRcolor for sequence super-resolution, spatiotemporal fusion, and generalization test for further performance test. [ABSTRACT FROM AUTHOR]

Radar, as one of the sensors for human activity recognition (HAR), has unique characteristics such as privacy protection and contactless sensing. Radar-based HAR has been applied in many fields such as human–computer interaction, smart surveillance and health assessment. Conventional machine learning approaches rely on heuristic hand-crafted feature extraction, and their generalization capability is limited. Additionally, extracting features manually is time–consuming and inefficient. Deep learning acts as a hierarchical approach to learn high-level features automatically and has achieved superior performance for HAR. This paper surveys deep learning based HAR in radar from three aspects: deep learning techniques, radar systems, and deep learning for radar-based HAR. Especially, we elaborate deep learning approaches designed for activity recognition in radar according to the dimension of radar returns (i.e., 1D, 2D and 3D echoes). Due to the difference of echo forms, corresponding deep learning approaches are different to fully exploit motion information. Experimental results have demonstrated the feasibility of applying deep learning for radar-based HAR in 1D, 2D and 3D echoes. Finally, we address some current research considerations and future opportunities. [ABSTRACT FROM AUTHOR]

Polarimetric synthetic aperture radar (PolSAR) has become increasingly popular in the past two decades, for it can derive multichannel features of ground objects, which contains more discriminative information compared with traditional SAR. In this paper, a neural nonlocal stacked sparse autoencoders with virtual adversarial regularization (NNSSAE-VAT) is proposed for PolSAR image classification. The NNSSAE first extracts the nonlocal features by calculating pairwise similarity of each pixel and its surrounding pixels using a neural network, which contains a multiscale feature extractor and a linear embedding layer. The feature extraction process can relieve the negative influence of speckle noise and extract discriminative nonlocal spatial information without carefully designed parameters. Then, the SSAE maps the center pixel and the extracted nonlocal features into deep latent space in which a Softmax classifier is utilized to conduct classification. The virtual adversarial training is introduced to regularize the network, which tries to keep the network from being overfitting. The experimental results from three real PolSAR image show that the proposed NNSSAE-VAT method has proved its robustness and effectiveness and it can achieve competitive performance compared with related methods. [ABSTRACT FROM AUTHOR]

In this paper, a component-based multi-layer parallel network is proposed for airplane detection in Synthetic Aperture Radar (SAR) imagery. In response to the problems called sparsity and diversity brought by SAR scattering mechanism, depth characteristics and component structure are utilized in the presented algorithm. Compared with traditional features, the depth characteristics have better description ability to deal with diversity. Component information is contributing in detecting complete targets. The proposed algorithm consists of two parallel networks and a constraint layer. First, the component information is introduced into the network by labeling. Then, the overall target and corresponding components are detected by the trained model. In the following discriminative constraint layer, the maximum probability and prior information are adopted to filter out wrong detection. Experiments for several comparative methods are conducted on TerraSAR-X SAR imagery; the results indicate that the proposed network has a higher accuracy for airplane detection. [ABSTRACT FROM AUTHOR]

Scene classification, aiming to identify the land-cover categories of remotely sensed image patches, is now a fundamental task in the remote sensing image analysis field. Deep-learning-model-based algorithms are widely applied in scene classification and achieve remarkable performance, but these high-level methods are computationally expensive and time-consuming. Consequently in this paper, we introduce a knowledge distillation framework, currently a mainstream model compression method, into remote sensing scene classification to improve the performance of smaller and shallower network models. Our knowledge distillation training method makes the high-temperature softmax output of a small and shallow student model match the large and deep teacher model. In our experiments, we evaluate knowledge distillation training method for remote sensing scene classification on four public datasets: AID dataset, UCMerced dataset, NWPU-RESISC dataset, and EuroSAT dataset. Results show that our proposed training method was effective and increased overall accuracy (3% in AID experiments, 5% in UCMerced experiments, 1% in NWPU-RESISC and EuroSAT experiments) for small and shallow models. We further explored the performance of the student model on small and unbalanced datasets. Our findings indicate that knowledge distillation can improve the performance of small network models on datasets with lower spatial resolution images, numerous categories, as well as fewer training samples. [ABSTRACT FROM AUTHOR]

This article describes the development of a machine learning (ML)-based algorithm for snowfall retrieval (Snow retrievaL ALgorithm fOr gpM–Cross Track, SLALOM-CT), exploiting ATMS radiometer measurements and using the CloudSat CPR snowfall products as references. During a preliminary analysis, different ML techniques (tree-based algorithms, shallow and convolutional neural networks—NNs) were intercompared. A large dataset (three years) of coincident observations from CPR and ATMS was used for training and testing the different techniques. The SLALOM-CT algorithm is based on four independent modules for the detection of snowfall and supercooled droplets, and for the estimation of snow water path and snowfall rate. Each module was designed by choosing the best-performing ML approach through model selection and optimization. While a convolutional NN was the most accurate for the snowfall detection module, a shallow NN was selected for all other modules. SLALOM-CT showed a high degree of consistency with CPR. Moreover, the results were almost independent of the background surface categorization and the observation angle. The reliability of the SLALOM-CT estimates was also highlighted by the good results obtained from a direct comparison with a reference algorithm (GPROF). [ABSTRACT FROM AUTHOR]

Fully Convolutional Networks (FCN) has shown better performance than other classifiers like Random Forest (RF), Support Vector Machine (SVM) and patch-based Deep Convolutional Neural Network (DCNN), for object-based classification using orthoimage only in previous studies; however, for further improving deep learning algorithm performance, multi-view data should be considered for training data enrichment, which has not been investigated for FCN. The present study developed a novel OBIA classification using FCN and multi-view data extracted from small Unmanned Aerial System (UAS) for mapping landcovers. Specifically, this study proposed three methods to automatically generate multi-view training samples from orthoimage training datasets to conduct multi-view object-based classification using FCN, and compared their performances with each other and also with RF, SVM, and DCNN classifiers. The first method does not consider the object surrounding information, while the other two utilized object context information. We demonstrated that all the three versions of FCN multi-view object-based classification outperformed their counterparts utilizing orthoimage data only. Furthermore, the results also showed that when multi-view training samples were prepared with consideration of object surroundings, FCN trained with these samples gave much better accuracy than FCN classification trained without context information. Similar accuracies were achieved from the two methods utilizing object surrounding information, although sample preparation was conducted using two different ways. When comparing FCN with RF, SVM, DCNN implies that FCN generally produced better accuracy than the other classifiers, regardless of using orthoimage or multi-view data. [ABSTRACT FROM AUTHOR]

Regional landslide identification is important for the risk management of landslide hazards. The traditional methods of regional landslide identification were mainly conducted by a human being. In previous studies, automatic landslide recognition mainly focused on new landslides distinct from the environment induced by rainfall or earthquake, using the image classification method and semantic segmentation method of deep learning. However, there is a lack of research on the automatic recognition of old loess landslides, which are difficult to distinguish from the environment. Therefore, this study uses the object detection method of deep learning to identify old loess landslides with Google Earth images. At first, a database of loess historical landslide samples was established for deep learning based on Google Earth images. A total of 6111 landslides were interpreted in three landslide areas in Gansu Province, China. Second, three object detection algorithms including the one-stage algorithm RetinaNet and YOLO v3 and the two-stage algorithm Mask R-CNN, were chosen for automatic landslide identification. Mask R-CNN achieved the greatest accuracy, with an AP of 18.9% and F1-score of 55.31%. Among the three landslide areas, the order of identification accuracy from high to low was Site 1, Site 2, and Site 3, with the F1-scores of 62.05%, 61.04% and 50.88%, respectively, which were positively related to their recognition difficulty. The research results proved that the object detection method can be employed for the automatic identification of loess landslides based on Google Earth images. [ABSTRACT FROM AUTHOR]

As satellite observation technology rapidly develops, the number of remote sensing (RS) images dramatically increases, and this leads RS image retrieval tasks to be more challenging in terms of speed and accuracy. Recently, an increasing number of researchers have turned their attention to this issue, as well as hashing algorithms, which map real-valued data onto a low-dimensional Hamming space and have been widely utilized to respond quickly to large-scale RS image search tasks. However, most existing hashing algorithms only emphasize preserving point-wise or pair-wise similarity, which may lead to an inferior approximate nearest neighbor (ANN) search result. To fix this problem, we propose a novel triplet ordinal cross entropy hashing (TOCEH). In TOCEH, to enhance the ability of preserving the ranking orders in different spaces, we establish a tensor graph representing the Euclidean triplet ordinal relationship among RS images and minimize the cross entropy between the probability distribution of the established Euclidean similarity graph and that of the Hamming triplet ordinal relation with the given binary code. During the training process, to avoid the non-deterministic polynomial (NP) hard problem, we utilize a continuous function instead of the discrete encoding process. Furthermore, we design a quantization objective function based on the principle of preserving triplet ordinal relation to minimize the loss caused by the continuous relaxation procedure. The comparative RS image retrieval experiments are conducted on three publicly available datasets, including UC Merced Land Use Dataset (UCMD), SAT-4 and SAT-6. The experimental results show that the proposed TOCEH algorithm outperforms many existing hashing algorithms in RS image retrieval tasks. [ABSTRACT FROM AUTHOR]

Synthetic aperture radar (SAR) imaging has developed rapidly in recent years. Although the traditional sparse optimization imaging algorithm has achieved effective results, its shortcomings are slow imaging speed, large number of parameters, and high computational complexity. To solve the above problems, an end-to-end SAR deep learning imaging algorithm is proposed. Based on the existing SAR sparse imaging algorithm, the SAR imaging model is first rewritten to the SAR complex signal form based on the real-value model. Second, instead of arranging the two-dimensional echo data into a vector to continuously construct an observation matrix, the algorithm only derives the neural network imaging model based on the iteration soft threshold algorithm (ISTA) sparse algorithm in the two-dimensional data domain, and then reconstructs the observation scene through the superposition and expansion of the multi-layer network. Finally, through the experiment of simulation data and measured data of the three targets, it is verified that our algorithm is superior to the traditional sparse algorithm in terms of imaging quality, imaging time, and the number of parameters. [ABSTRACT FROM AUTHOR]

When original aerial photographs are combined with deep learning to classify forest vegetation cover, these photographs are often hindered by the interlaced composition of complex backgrounds and vegetation types as well as the influence of different deep learning calculation processes, resulting in unpredictable training and test results. The purpose of this research is to evaluate (1) data preprocessing, (2) the number of classification targets, and (3) convolutional neural network (CNN) approaches combined with deep learning's effects on high-resolution aerial photographs to identify forest and vegetation types. Data preprocessing is mainly composed of principal component analysis and content simplification (noise elimination). The number of classification targets is divided into 14 types of forest vegetation that are more complex and difficult to distinguish and seven types of forest vegetation that are simpler. We used CNN approaches to compare three CNN architectures: VGG19, ResNet50, and SegNet. This study found that the models had the best execution efficiency and classification accuracy after data preprocessing using principal component analysis. However, an increase in the number of classification targets significantly reduced the classification accuracy. The algorithm analysis showed that VGG19 achieved the best classification accuracy, but SegNet achieved the best performance and overall stability of relative convergence. This proves that data preprocessing helps identify forest and plant categories in aerial photographs with complex backgrounds. If combined with the appropriate CNN algorithm, these architectures will have great potential to replace high-cost on-site forestland surveys. At the end of this study, a user-friendly classification system for practical application is proposed, and its testing showed good results. [ABSTRACT FROM AUTHOR]

Existing airborne SAR autofocus methods can be classified as parametric and non-parametric. Generally, non-parametric methods, such as the widely used phase gradient autofocus (PGA) algorithm, are only suitable for scenes with many dominant point targets, while the parametric ones are suitable for all types of scenes, in theory, but their efficiency is generally low. In practice, whether many dominant point targets are present in the scene is usually unknown, so determining what kind of algorithm should be selected is not straightforward. To solve this issue, this article proposes an airborne SAR autofocus approach combined with blurry imagery classification to improve the autofocus efficiency for ensuring autofocus precision. In this approach, we embed the blurry imagery classification based on a typical VGGNet in a deep learning community into the traditional autofocus framework as a preprocessing step before autofocus processing to analyze whether dominant point targets are present in the scene. If many dominant point targets are present in the scene, the non-parametric method is used for autofocus processing. Otherwise, the parametric one is adopted. Therefore, the advantage of the proposed approach is the automatic batch processing of all kinds of airborne measured data. [ABSTRACT FROM AUTHOR]

In remote sensing images, small target size and diverse background cause difficulty in locating targets accurately and quickly. To address the lack of accuracy and inefficient real-time performance of existing tracking algorithms, a multi-object tracking (MOT) algorithm for ships using deep learning was proposed in this study. The feature extraction capability of target detectors determines the performance of MOT algorithms. Therefore, you only look once (YOLO)-v3 model, which has better accuracy and speed than other algorithms, was selected as the target detection framework. The high similarity of ship targets will cause poor tracking results; therefore, we used the multiple granularity network (MGN) to extract richer target appearance information to improve the generalization ability of similar images. We compared the proposed algorithm with other state-of-the-art multi-object tracking algorithms. Results show that the tracking accuracy is improved by 2.23%, while the average running speed is close to 21 frames per second, meeting the needs of real-time tracking. [ABSTRACT FROM AUTHOR]