Search

The identification and early warning of potential landslides can effectively reduce the number of casualties and the amount of property loss. At present, interferometric synthetic aperture radar (InSAR) is considered one of the mainstream methods for the large-scale identification and detection of potential landslides, and it can obtain long-term time-series surface deformation data. However, the method of identifying anomalous deformation areas using InSAR data is still mainly manual delineation, which is time-consuming, labor-consuming, and has no generally accepted criterion. In this study, a two-stage detection deep learning network (InSARNet) is proposed and used to detect anomalous deformation areas in Maoxian County, Sichuan Province. Compared with the most commonly used detection models, it is demonstrated that the InSARNet has a better performance in the detection of anomalous deformation in mountainous areas, and all of the quantitative evaluation indexes are higher for InSARNet than for the other models. After the anomalous deformation areas are identified using the proposed model, the possible relationship between the anomalous deformation areas and potential landslides is investigated. Finally, the fact that the automatic and rapid identification of potential landslides is the inevitable trend of future development is discussed. [ABSTRACT FROM AUTHOR]

Both short-distance association knowledge and long-distance interaction knowledge in the knowledge base contain rich semantics. When learning entity and relation representation in a knowledge base, if we can learn short-distance association knowledge and long-distance interaction knowledge at the same time, we can learn the representation method with rich semantic information and keep the original structure of the knowledge base. Among the knowledge contained in a large number of records, some knowledge reflects individual characteristics and can be called local knowledge; others reflect group characteristics and can be called global knowledge. Using different ways to learn local and global knowledge in the deep learning model will better reflect the difference between the two kinds of knowledge at the model level, and make the model have the ability to understand both individual characteristics and overall characteristics. Through layer-by-layer forward propagation and error back propagation algorithms, the entire network is gradually optimized in an "end-to-end" manner. This "end-to-end" approach leaves some means of introducing prior knowledge flexibly into the model. Although it can reduce the burden on researchers, this "data-driven" approach brings the shortcomings of poor interpretability of learning results and weak generalization ability. Combining the specific prior knowledge implicit in the data with the deep learning algorithm can optimize the algorithm in a targeted manner and avoid blind searching in the solution space, so as to obtain a model with better performance and wider use. To this end, this paper investigates combining prior knowledge with deep learning to design efficient algorithms to address the classification of English translation styles. This paper combines local knowledge with global knowledge and deep learning methods and proposes a memory neural network method combining local knowledge and global knowledge. By recording the local knowledge in the local memory module and simultaneously recording the global knowledge in the global memory module, the method effectively learns the latent information in a large number of records. This paper combines short-distance association knowledge with long-distance interaction knowledge and a distributed representation learning method based on deep learning and proposes a deep learning method combining short-distance association knowledge and long-distance interaction knowledge. On the IWSLT English translation task, experiments show that the method significantly improves translation quality, confirming that grammatical dependencies enhance attention by supplementing dependent grammatical information, resulting in more effective and richer context vectors that more accurately represent contextual situations. Additional experimental analysis showed that the model underwent careful parameter selection and analysis. By mining valuable long-distance interactive knowledge in the knowledge base and using it in the distributed representation learning of the knowledge base, while constraining the short-distance related knowledge and constraining the long-distance interactive knowledge, the learned knowledge can be used to effectively complete the knowledge base distributed representation for discovering new relations. [ABSTRACT FROM AUTHOR]

• A staged data preprocessing strategy for honey peach young fruit is designed. • The SCAE is proposed to ensure the discriminability of the extracted features. • The SCAE is embedded into the DeepSORT to form the DeepSORT-SCAE approach. • The DeepSORT-SCAE can measure the spatially continuous trajectory of the young fruit. • The effectiveness of the developed method is verified by experiments. The young fruit bagging robot usually preferentially selects the object with largest prediction box size for bagging. However, due to the uncertainty in the output of the traditional models, the expected endpoint of mechanical motion may change constantly, which affects the bagging efficiency. This paper proposes a novel tracking algorithm for honey peach young fruit targets. Firstly, according to the proportion of young fruit prediction box in the original image, a staged data preprocessing strategy is designed. Then, the siamese convolution autoencoder (SCAE) is developed by combining autoencoding mechanism with siamese training mode, and the SCAE is trained in two-stages to ensure the discriminability of the extracted appearance features. Finally, the SCAE is inserted into the DeepSORT architecture to generate the DeepSORT-SCAE object tracking algorithm, which can measure the spatially continuous trajectory of multiple young fruit targets within video streams. The effectiveness of the developed method is verified by experiments. [ABSTRACT FROM AUTHOR]

Deep learning-based fault diagnosis methods have to be trained by a large amount of labeled data for accurate diagnosis results. However, in real-case industrial scenarios, the data available is limited since the difficulty of obtaining fault samples. In addition, the monitoring data always come from various working conditions. As a result, data scarcity and domain shift have become two main factors that limit the performance of diagnosis models. We called this problem the few-shot cross-domain fault diagnosis, which has been rarely explored. To deal with the above challenge, we propose a novel uncertainty-based contrastive prototype-matching network (UCPMnet). Specifically, two main parts are conducted for UCPMnet. Considering the construction of representation spaces for compact intra-class and sparse inter-class embeddings, the contrastive prototypical network (CPN) is established by remolding the prototypical network. Subsequently, an uncertainty-based prototype-matching (UPM) method is presented for class-wise domain adaptation based on pseudo-label learning. At the same time, an estimation approach of prediction uncertainty is introduced to adaptively rectify the threshold for screening the pseudo label, which aims to mitigate the effect of noisy predictions for prototype-matching. Extensive experiments are conducted on two datasets, and four classical methods are selected for comparison. The results corroborate the superiority of the UCPMnet for fault diagnosis in the case of small data. [ABSTRACT FROM AUTHOR]

[Display omitted] • The attention mechanism implements adaptive feature non-linear transformation. • A hyperparameter adaptive activation function is developed based on tanh function. • Designed activation function avoids non-zero mean and gradient vanishing problems. • A deep adaptive network based on ResNet is proposed for bearing fault diagnosis. • Comparative experiments verify the superiority of proposed fault diagnosis model. Bearing health condition directly affects the reliability of mechanical equipment. Although deep learning (DL) algorithms have achieved great results in the field of bearing fault diagnosis, traditional activation function uses a fixed mathematical formula to achieve non-linear feature transformation, which tend to compress part of the effective fault information and reduce the performance of fault diagnosis. To address this problem, this paper proposes a slope and threshold adaptive activation function with tanh function (STAC-tanh). Establish the relationship between non-linear feature transformation and input signal by automatically adjusting the shape of activation function. Finally, the model can retain valid fault information to improve fault diagnosis performance. Then, combining STAC-tanh and Residual Networks, this paper proposes ResNet-STAC-tanh for bearing fault diagnosis. Experimenting on the two bearing datasets with added noise, the average accuracy of the network reached 90.00% and 90.77%, respectively. The effectiveness of the new method was verified through comparative experiments. [ABSTRACT FROM AUTHOR]

Optical information storage, in virtue of its large capacity, high stability, and long longevity, holds promising prospects in mass storage, while being limited by the trade‐off between readout quality and error rate. The emerging intersection of optical storage and deep learning presents a valuable opportunity to achieve high‐fidelity data storage. Here, a novel paradigm of error‐free long‐lifespan optical storage enhanced is proposed by deep learning, harnessing neural network to extract optical information from birefringence measurements. It is demonstrated that using neural networks outperforms traditional approaches in terms of efficiency and accuracy. Moreover, by adding extra birefringence information as input to the neural network, nearly 100%$100\%$ accuracy is achieved on an established five‐bit dataset. Remarkably, even under extremely severe ambiguity, the paradigm still fulfills error‐free readout and maintains a long lifespan. The experimental storage scheme is significantly conducive to the development of large‐scale error‐free storage, and paves the way for robust optical storage with environmental and temporal tolerance in practical scenarios. [ABSTRACT FROM AUTHOR]

Ultrasound has become an indispensable clinical scanning and diagnostic tool because of its convenience, rapidity, and radiation-free properties. In ultrasound imaging, the resolution and detection depth of ultrasound images are closely related to the frequency of the transmitted ultrasound. Ultrasound with low transmission frequency has a wide detection range, but its images have problems such as low resolution, blurred edge contours and obvious background noise. In contrast, the higher transmission frequency, the smaller the detection range, but the resolution has improved. The rapid development of deep learning provides new approaches and enormous potential for improving resolution of ultrasound images and acquiring high-resolution images with an extensive detection range. In this paper, the attention mechanism is combined into a convolutional neural network based on the encoder-decoder, and the feature maps extracted at different scales are merged by skip connections to preserve more structure and contrast details, thus realizing the improvement of lower-frequency ultrasound image resolution. This work utilizes the capacitive micromechanical ultrasound transducer (CMUT) probe for dataset acquisition. The network's performance is validated by simulated datasets for point targets and experimental datasets for breast body models. The experimental results show that for ultrasound images, the network output map can improve by 0.2 in structural similarity and at least 11 dB in peak signal-to-noise ratio. The proposed network can effectively preserve ultrasound images' structural and textural information, improve edge contour definition, and significantly suppress noise and imaging artifacts, which has significant development potential in ultrasound imaging. [Display omitted] • Different ultrasound emission frequencies have significant impact on imaging quality. • Datasets of different frequencies using capacitive micromechanical ultrasound transducer (CMUT) probes. • Using deep learning to improve the quality of low-frequency images by extracting features from high-frequency images. • With incorporation of the attention mechanism, the network highlights the prominent features through skip connections while maintaining computational efficiency. [ABSTRACT FROM AUTHOR]

Agricultural drought seriously threatens the food and ecological security of most of the world's developing countries. Data-driven integrated agricultural drought index with remote sensing provides an effective tool to monitor, evaluate, and predict the agricultural drought. However, there is still a lack of comprehensive analytical work on taking the most effective machine learning (ML) and deep learning (DL) methods to construct such integrated drought index. In other words, it is still unclear whether the recent DL methods can improve integrated drought monitoring as compared with the currently widely used ML methods. Therefore, we critically evaluated the performances of four representative DL methods (represents the four currently popular DL network types) i.e., Entity Embedding Deep Neural Network (EEDNN), One-dimensional Convolutional Neural Network (1D-CNN), Gated Recurrent Unit (GRU), and Self-Attention Mechanism (SAM) and three widely used tree-based ML methods i.e., Cubist, Random Forest (RF), and Light Gradient Boosting Machine (LGBM), through constructing a QuickDRI like integrated drought index (abbreviated as QuickDRI-China). About 30 years of meteorological data, 14 years of remote sensing data, and various biophysical variables in China such as land use/land cover, available water capacity, irrigated agriculture, elevation, and ecoregion were employed in this study. Results showed that the EEDNN performed best, followed by the RF and LGBM, and then the other methods including the currently wide used Cubist, according to the station accuracy evaluations, spatial description evaluations, and responses to specific drought event. The tree-based ML methods such as RF and LGBM are still competitive in constructing the integrated agricultural drought index at the current stage. However, the higher accuracy, the smoother spatial description, and the more responsive ability of the EEDNN demonstrate great potential of DL methods. The future integrated agricultural drought monitoring with remote sensing should develop a specialized DL network for heterogeneous agricultural drought features. • Various deep learning approaches were examined for integrated drought monitoring. • Tree-based ensemble learning methods were effective. • Deep learning has unique advantages for assessing spatial description of agricultural drought. • Deep learning has great potential to improve integrated drought monitoring. [ABSTRACT FROM AUTHOR]

Early blight and late blight are important factors restricting tomato yield. However, it is still a challenge to accurately and objectively detect and segment crop diseases in order to evaluate disease damage. In this paper, the Disease Segmentation Detection Transformer (DS-DETR) is proposed to segment leaf disease spots efficiently based on several improvements to DETR. Additionally, a damage assessment is carried out by the area ratio of the segmented leaves to the disease spots. First, an unsupervised pre-training method was introduced into DETR with the Plant Disease Classification Dataset (PDCD) to solve the problem of the long training epochs and slow convergence speed of DETR. This method can train the Transformer structures in advance to obtain leaf disease features. Loading the pre-training model weight in DS-DETR can speed up the convergence speed of the model. Then, Spatially Modulated Co-Attention (SMCA) was used to assign Gaussian-like spatial weights to the query box of DS-DETR. The different positions in the image are trained using the query boxes with different weights to improve the accuracy of the model. Finally, an improved relative position code was added to the Transformer structure of DS-DETR. Relative position coding promotes the capture of the sequence order of input tokens by the Transformer. The spatial location feature is strengthened by establishing the location relationship between different instances. Based on these improvements, the DS-DETR model was tested on the Tomato leaf Disease Segmentation Dataset (TDSD) constructed by us. The experimental results show that the DS-DETR proposed by us achieved 0.6823 for APmask, which improved by 12.87%, 8.25%, 3.67%, 1.95%, 10.27%, and 9.52% compared with the state-of-the-art: Mask RCNN, BlendMask, CondInst, SOLOv2, ISTR, and DETR, respectively. In addition, the disease grading accuracy reached 0.9640 according to the segmentation results given by our proposed model. [ABSTRACT FROM AUTHOR]

In image denoising, the recovery of high-frequency regions, such as image edges, directly affects the quality of the denoised images. However, previous deep learning-based denoising methods fail to effectively allocate the transmission of different frequency information and have difficulty giving network attention to high-frequency regions. In this paper, we rethink the fusion of image gradients in a neural network and deeply mine the intrinsic structure of the input image to propose a novel layered input gradient network (LIGN) for image denoising. The core of our network focuses on the features of different frequencies through two networks, which contain several key elements: (a) The input noise image is layered to widen the shallow layer of the network and to promote the hierarchical learning of different types of frequencies. (b) A multiscale feature extraction (MFE) block and information shunting (IS) block are proposed to integrate and separate various frequency features. (c) A gradient network (GradiNet) is designed to extract high-frequency information by network training, and the information is adaptively added to the input of the parallel main network (MainNet) through normalization to obtain high-quality images. Furthermore, we propose a sharpening loss function to enhance the texture details of the denoised image and improve visual quality. Extensive experiments on synthetic and real-world datasets show that the proposed method greatly enhances perceptual visual quality and achieves state-of-the-art performance on both PSNR and SSIM. The source code and pretrained models are available at https://github.com/JerryYann/LIGN. • A layered input gradient network (LIGN) based on a dual U-Net for high-quality image denoising is proposed. • Layered input and sharpening loss greatly improve the perceptual quality of the denoised image. • Multi-scale feature extraction block can capture more semantic information. • LIGN achieves the SoTA performance compared with the latest methods on synthetic and real noise datasets. [ABSTRACT FROM AUTHOR]

Pansharpening is an important yet challenging remote sensing image processing task, which aims to reconstruct a high-resolution (HR) multispectral (MS) image by fusing a HR panchromatic (PAN) image and a low-resolution (LR) MS image. Though deep learning (DL)-based pansharpening methods have achieved encouraging performance, they are infeasible to fully utilize the deep semantic features and shallow contextual features in the process of feature fusion for a HR-PAN image and LR-MS image. In this paper, we propose an efficient full-depth feature fusion network (FDFNet) for remote sensing pansharpening. Specifically, we design three distinctive branches called PAN-branch, MS-branch, and fusion-branch, respectively. The features extracted from the PAN and MS branches will be progressively injected into the fusion branch at every different depth to make the information fusion more broad and comprehensive. With this structure, the low-level contextual features and high-level semantic features can be characterized and integrated adequately. Extensive experiments on reduced- and full-resolution datasets acquired from WorldView-3, QuickBird, and GaoFen-2 sensors demonstrate that the proposed FDFNet only with less than 100,000 parameters performs better than other detail injection-based proposals and several state-of-the-art approaches, both visually and quantitatively. [ABSTRACT FROM AUTHOR]

Lamb wave-based detection has become a promising technique for structural health monitoring in plate-like structures. However, the dispersion effect of Lamb waves makes the wave packets elongated, which degrades the resolution for damage identification. Hence, it is necessary to develop effective methods for the dispersion removal of Lamb waves. Recently, deep learning has drawn much attention for Lamb wave detection due to its powerful feature extraction capability. Most current deep learning models are developed directly for damage classification or location estimation tasks on the Lamb wave signals, rather than signal processing for further manipulation. Therefore, in this paper, a novel approach based on a convolutional auto-encoder is proposed for the dispersion compensation of Lamb waves. The convolutional auto-encoder model is utilized to construct the mapping relationship between the dispersive signal and the time of flight of the wave packet. The dispersion compensated signal is reconstructed by combining the estimated time of flight from the network with the waveform of the excitation signal. Numerical and experimental validations on aluminum and composite plates are implemented to verify the effectiveness of the proposed method. Compared with conventional methods for dispersion compensation, the results demonstrate that our proposed method not only separates the overlapped wave packets but also enables dispersion compensation of multimodal and multi-packet Lamb wave signals. In addition, this method is still applicable in the case of inaccurate dispersion data due to the generalizability of the data-driven model. [ABSTRACT FROM AUTHOR]

• A study on the use of deep learning model in the field of neutron image restoration is performed. • A lightweight, convenient and efficient deep learning network is proposed for neutron image restoration. • The X-ray dataset solves the problem of lack of real neutron image datasets. • The pre-training model achieves good results in real neutron image restoration. Traditional neutron image restoration (NIR) methods lack flexibility and generalization in dealing with multiple image restoration tasks, which indirectly limits the practical applications of neutron images. Deep convolutional neural networks (CNNs) have achieved impressive success in image restoration, so this paper preliminarily explores and applies deep CNNs in NIR. Considering the characteristics and applications of neutron images, we propose a fast and lightweight densely connected attention U-Net (DAUNet) for NIR. We strengthen the network's attention to the image structural information and the processing at different frequencies by adding an attention mechanism to the symmetric skip connection under each scale of U-Net. The proposed DAUNet uses a single model for recovering images with different degradations, such as additive noise and blur. Extensive experiments on synthetic and real neutron-degraded images are conducted, and the results show that DAUNet is effective and efficient in NIR and achieves satisfactory results in terms of both comprehensive performance (PSNR) and intrinsic metrics (parameters, memory cost and running time), which are highly attractive for practical applications. In addition, to address the problem of lacking real NIR datasets, we use X-ray images that the imaging principles are closer to that of neutron imaging as data drivers for the network. [ABSTRACT FROM AUTHOR]

Data-driven deep learning approaches have been recently developed for guided wave-based structural health monitoring. However, the difficulty in collecting and labeling valid samples often leads to a small dataset in practical damage identification, lowering the performance of the trained model. In addition, conventional deep learning models often lack a certain degree of physical interpretability. In this study, an attention-based interpretable prototypical network is proposed for small-sample damage identification using ultrasonic guided waves. The prototypical network is utilized as the framework to calculate the prototype of each category and the similarity between samples based on metric space. Afterward, the channel attention module is constructed for feature extraction, which enables the network to highlight valid information across channels and alleviate the overfitting problem. Moreover, local interpretable model-agnostic explanation (LIME) is introduced to explain the intrinsic mechanism for damage identification performed by the network in terms of critical feature contributions. To implement efficient damage identification based on small data, both numerical and experimental studies are carried out, in which the dataset contains pinhole, crack, and corrosion damage. The classification performance shows that the proposed network can serve as an effective model to overcome the shortage of limited data for damage identification, and the LIME analysis significantly enhances the interpretability of the network. [ABSTRACT FROM AUTHOR]