Search

As globalization accelerates, the linguistic diversity and semantic complexity of in-vehicle communication is increasing. In order to meet the needs of different language speakers, this paper proposes an interactive attention-based contrastive learning framework (IABCL) for the field of in-vehicle dialogue, aiming to effectively enhance cross-lingual natural language understanding (NLU). The proposed framework aims to address the challenges of cross-lingual interaction in in-vehicle dialogue systems and provide an effective solution. IABCL is based on a contrastive learning and attention mechanism. First, contrastive learning is applied in the encoder stage. Positive and negative samples are used to allow the model to learn different linguistic expressions of similar meanings. Its main role is to improve the cross-lingual learning ability of the model. Second, the attention mechanism is applied in the decoder stage. By articulating slots and intents with each other, it allows the model to learn the relationship between the two, thus improving the ability of natural language understanding in languages of the same language family. In addition, this paper constructed a multilingual in-vehicle dialogue (MIvD) dataset for experimental evaluation to demonstrate the effectiveness and accuracy of the IABCL framework in cross-lingual dialogue. With the framework studied in this paper, IABCL improves by 2.42% in intent, 1.43% in slot, and 2.67% in overall when compared with the latest model. [ABSTRACT FROM AUTHOR]

As one of the advanced automotive chassis technologies, the steer-by-wire system offers a high level of precision, responsiveness, and controllability in the driving experience. It can also adjust and optimize parameters to adapt to the preferences of different drivers. However, when faced with the steer-by-wire system, both experienced drivers and novice drivers are in the novice stage, exhibiting learning or adaptation behaviors when using this steering system. In this paper, a small-scale pilot evaluation was conducted by means of a questionnaire survey and driving-simulator experiment, and the learning behavior and adaptability of four experienced and four novice drivers to the steer-by-wire system were analyzed when using the traditional steering system. The study found that experienced drivers show significant changes in their adaptation to the steering system, mainly due to their habitual driving with traditional steering systems. In contrast, novice drivers show no significant changes in their adaptation to the steering system, which is attributed to their lack of driving experience and skills, resulting in less sensitivity to changes in the steering system. Additionally, the study found that novice drivers under the steer-by-wire system grasp control over speed and steering-wheel angle more quickly. This research provides a reference for improving drivers' learning and adaptation abilities to the steer-by-wire system and optimizing the design of the steer-by-wire system. [ABSTRACT FROM AUTHOR]

The primary sight control system of a tank gunner has image stabilization as one of its primary functions. The image stabilization deviation in the aiming line is a key indicator for evaluating the operational status of Gunner's Primary Sight control system. Employing image detection technology to measure image stabilization deviation enhances the effectiveness and accuracy of the detection process and allows for the evaluation of image stabilization functionality. Hence, this paper proposes an image detection method aimed at the Gunner's Primary Sight control system of a specific tank which utilizes an enhanced You Only Look Once version 5 (YOLOv5) sight-stabilizing deviation algorithm. At first, a dynamic weight factor is integrated into SCYLLA-IoU (SIOU), creating δ-SIOU, which replaces Complete IoU (CIoU) as the loss function of YOLOv5. After that, the Spatial Pyramid Pool module of YOLOv5 was enhanced to improve the multi-scale feature fusion ability of the model, thereby elevating the performance of the detection model. Finally, the C3CA module was created by embedding the Coordinate Attention (CA) attention mechanism into the CSK-MOD-C3 (C3) module. The Bi-directional Feature Pyramid (BiFPN) network structure was also incorporated into the Neck network of YOLOv5 to improve the model's ability to learn target location information and image detection accuracy. Based on data collected by a mirror control test platform, experimental results indicate an improvement in the detection accuracy of the model by 2.1%. These findings offer valuable insights into measuring the image stabilization deviation in the aiming line and facilitating the development of the parameter measurement system for Gunner's Primary Sight control system. [ABSTRACT FROM AUTHOR]

In response to the issues of low model recognition accuracy and weak generalization in mechanical equipment fault diagnosis due to scarce data, this paper proposes an innovative solution, a cross-device secondary transfer-learning method based on EGRUN (efficient gated recurrent unit network). This method utilizes continuous wavelet transform (CWT) to transform source domain data into images. The EGRUN model is initially trained, and shallow layer weights are frozen. Subsequently, random overlapping sampling is applied to the target domain data to enhance data and perform secondary transfer learning. The experimental results demonstrate that this method not only significantly improves the model's ability to learn fault features but also enhances its classification accuracy and generalization performance. Compared to current state-of-the-art algorithms, the model proposed in this study shows faster convergence speed, higher diagnostic accuracy, and superior robustness and generalization, providing an effective approach to address the challenges arising from scarce data and varying operating conditions in practical engineering scenarios. [ABSTRACT FROM AUTHOR]

Dongba characters are ancient ideographic scripts with abstract expressions that differ greatly from modern Chinese characters; directly applying existing methods cannot achieve the font style transfer of Dongba characters. This paper proposes an Attention-based Font style transfer Generative Adversarial Network (AFGAN) method. Based on the characteristics of Dongba character images, two core modules are set up in the proposed AFGAN, namely void constraint and font stroke constraint. In addition, in order to enhance the feature learning ability of the network and improve the style transfer effect, the Convolutional Block Attention Module (CBAM) mechanism is added in the down-sampling stage to help the network better adapt to input font images with different styles. The quantitative and qualitative analyses of the generated font and the real font were conducted by consulting with professional artists based on the newly built small seal script, slender gold script, and Dongba character dataset, and the styles of the small seal script and slender gold script were transferred to Dongba characters. The results indicate that the proposed AFGAN method has advantages in evaluation indexes and visual quality compared to existing networks. At the same time, this method can effectively learn the style features of small seal script and slender gold script, and transfer them to Dongba characters, indicating the effectiveness of this method. [ABSTRACT FROM AUTHOR]

In order to efficiently identify early tea diseases, an improved YOLOv8 lesion detection method is proposed to address the challenges posed by the complex background of tea diseases, difficulty in detecting small lesions, and low recognition rate of similar phenotypic symptoms. This method focuses on detecting tea leaf blight, tea white spot, tea sooty leaf disease, and tea ring spot as the research objects. This paper presents an enhancement to the YOLOv8 network framework by introducing the Receptive Field Concentration-Based Attention Module (RFCBAM) into the backbone network to replace C2f, thereby improving feature extraction capabilities. Additionally, a mixed pooling module (Mixed Pooling SPPF, MixSPPF) is proposed to enhance information blending between features at different levels. In the neck network, the RepGFPN module replaces the C2f module to further enhance feature extraction. The Dynamic Head module is embedded in the detection head part, applying multiple attention mechanisms to improve multi-scale spatial location and multi-task perception capabilities. The inner-IoU loss function is used to replace the original CIoU, improving learning ability for small lesion samples. Furthermore, the AKConv block replaces the traditional convolution Conv block to allow for the arbitrary sampling of targets of various sizes, reducing model parameters and enhancing disease detection. the experimental results using a self-built dataset demonstrate that the enhanced YOLOv8-RMDA exhibits superior detection capabilities in detecting small target disease areas, achieving an average accuracy of 93.04% in identifying early tea lesions. When compared to Faster R-CNN, MobileNetV2, and SSD, the average precision rates of YOLOv5, YOLOv7, and YOLOv8 have shown improvements of 20.41%, 17.92%, 12.18%, 12.18%, 10.85%, 7.32%, and 5.97%, respectively. Additionally, the recall rate (R) has increased by 15.25% compared to the lowest-performing Faster R-CNN model and by 8.15% compared to the top-performing YOLOv8 model. With an FPS of 132, YOLOv8-RMDA meets the requirements for real-time detection, enabling the swift and accurate identification of early tea diseases. This advancement presents a valuable approach for enhancing the ecological tea industry in Yunnan, ensuring its healthy development. [ABSTRACT FROM AUTHOR]

Fitness is important in people's lives. Good fitness habits can improve cardiopulmonary capacity, increase concentration, prevent obesity, and effectively reduce the risk of death. Home fitness does not require large equipment but uses dumbbells, yoga mats, and horizontal bars to complete fitness exercises and can effectively avoid contact with people, so it is deeply loved by people. People who work out at home use social media to obtain fitness knowledge, but learning ability is limited. Incomplete fitness is likely to lead to injury, and a cheap, timely, and accurate fitness detection system can reduce the risk of fitness injuries and can effectively improve people's fitness awareness. In the past, many studies have engaged in the detection of fitness movements, among which the detection of fitness movements based on wearable devices, body nodes, and image deep learning has achieved better performance. However, a wearable device cannot detect a variety of fitness movements, may hinder the exercise of the fitness user, and has a high cost. Both body-node-based and image-deep-learning-based methods have lower costs, but each has some drawbacks. Therefore, this paper used a method based on deep transfer learning to establish a fitness database. After that, a deep neural network was trained to detect the type and completeness of fitness movements. We used Yolov4 and Mediapipe to instantly detect fitness movements and stored the 1D fitness signal of movement to build a database. Finally, MLP was used to classify the 1D signal waveform of fitness. In the performance of the classification of fitness movement types, the mAP was 99.71%, accuracy was 98.56%, precision was 97.9%, recall was 98.56%, and the F1-score was 98.23%, which is quite a high performance. In the performance of fitness movement completeness classification, accuracy was 92.84%, precision was 92.85, recall was 92.84%, and the F1-score was 92.83%. The average FPS in detection was 17.5. Experimental results show that our method achieves higher accuracy compared to other methods. [ABSTRACT FROM AUTHOR]

This paper introduces an FPGA-based implementation of a smart switch designed to avoid inrush currents occurring during the connection of single-phase transformers utilized in grid-connected photovoltaic (PV) systems. The magnitude of inrush currents is notably impacted by the residual flux within the transformer core and the precise moment of energization relative to the wave cycle. Alternative methods frequently hinge on intricate procedures to estimate residual flux. This challenge is adeptly circumvented by the innovative smart control system proposed herein, rendering it a cost-effective solution for grid-connected PV systems. The proposed solution for mitigating inrush current remains effective, even in the face of challenges with current and voltage sensors. This resilience arises from the system's ability to learn and adapt by leveraging information acquired from the network. [ABSTRACT FROM AUTHOR]

Although convolutional neural networks (CNNs) have produced great achievements in various fields, many scholars are still exploring better network models, since CNNs have an inherent limitation—that is, the remote modeling ability of convolutional kernels is limited. On the contrary, the transformer has been applied by many scholars to the field of vision, and although it has a strong global modeling capability, its close-range modeling capability is mediocre. While the foreground information to be segmented in medical images is usually clustered in a small interval in the image, the distance between different categories of foreground information is uncertain. Therefore, in order to obtain a perfect medical segmentation prediction graph, the network should not only have a strong learning ability for local details, but also have a certain distance modeling ability. To solve these problems, a remote feature exploration (RFE) module is proposed in this paper. The most important feature of this module is that remote elements can be used to assist in the generation of local features. In addition, in order to better verify the feasibility of the innovation in this paper, a new multi-organ segmentation dataset (MOD) was manually created. While both the MOD and Synapse datasets label eight categories of organs, there are some images in the Synapse dataset that label only a few categories of organs. The proposed method achieved 79.77% and 75.12% DSC on the Synapse and MOD datasets, respectively. Meanwhile, the HD95 (mm) scores were 21.75 on Synapse and 7.43 on the MOD dataset. [ABSTRACT FROM AUTHOR]

Object detection based on wood defects involves using bounding boxes to label defects in the surface image of the wood. This step is crucial before the transformation of wood products. Due to the small size and diverse shape of wood defects, most previous object detection models are unable to filter out critical features effectively. Consequently, they have faced challenges in generating adequate contextual information to detect defects accurately. In this paper, we proposed a YOLOv5 model based on a Semi-Global Network (SGN) to detect wood defects. Unlike previous models, firstly, a lightweight SGN is introduced in the backbone to model the global context, which can improve the accuracy and reduce the complexity of the network at the same time; the backbone is embedded with the Extended Efficient Layer Aggregation Network (E-ELAN), which continuously enhances the learning ability of the network; and finally, the Efficient Intersection and Merger (EIOU) loss is used to solve the problems of slow convergence speed and inaccurate regression results. Experimental results on public wood defect datasets demonstrated that our approach outperformed existing target detection models. The mAP value was 86.4%, a 3.1% improvement over the baseline network model, a 7.1% improvement over SSD, and a 13.6% improvement over Faster R-CNN. These results show the effectiveness of our proposed methodology. [ABSTRACT FROM AUTHOR]

Anomaly detection tasks involving time-series signal processing have been important research topics for decades. In many real-world anomaly detection applications, no specific distributions fit the data, and the characteristics of anomalies are different. Under these circumstances, the detection algorithm requires excellent learning ability of the data features. Transformers, which apply the self-attention mechanism, have shown outstanding performances in modelling long-range dependencies. Although Transformer based models have good prediction performance, they may be influenced by noise and ignore some unusual details, which are significant for anomaly detection. In this paper, a novel temporal context fusion framework: Temporal Context Fusion Transformer (TCF-Trans), is proposed for anomaly detection tasks with applications to time series. The original feature transmitting structure in the decoder of Informer is replaced with the proposed feature fusion decoder to fully utilise the features extracted from shallow and deep decoder layers. This strategy prevents the decoder from missing unusual anomaly details while maintaining robustness from noises inside the data. Besides, we propose the temporal context fusion module to adaptively fuse the generated auxiliary predictions. Extensive experiments on public and collected transportation datasets validate that the proposed framework is effective for anomaly detection in time series. Additionally, the ablation study and a series of parameter sensitivity experiments show that the proposed method maintains high performance under various experimental settings. [ABSTRACT FROM AUTHOR]

Defect detection in steel surface focuses on accurately identifying and precisely locating defects on the surface of steel materials. Methods of defect detection with deep learning have gained significant attention in research. Existing algorithms can achieve satisfactory results, but the accuracy of defect detection still needs to be improved. Aiming at this issue, a hybrid attention network is proposed in this paper. Firstly, a CBAM attention module is used to enhance the model's ability to learn effective features. Secondly, an adaptively spatial feature fusion (ASFF) module is used to improve the accuracy by extracting multi-scale information of defects. Finally, the CIOU algorithm is introduced to optimize the training loss of the baseline model. The experimental results show that the performance of our method in this work is superior on the NEU-DET dataset, with an 8.34% improvement in mAP. Compared with major algorithms of object detection such as SSD, EfficientNet, YOLOV3, and YOLOV5, the mAP was improved by 16.36%, 41.68%, 20.79%, and 13.96%, respectively. This demonstrates that the mAP of our proposed method is higher than other major algorithms. [ABSTRACT FROM AUTHOR]

Defect detection in power scenarios is a critical task that plays a significant role in ensuring the safety, reliability, and efficiency of power systems. The existing technology requires enhancement in its learning ability from large volumes of data to achieve ideal detection effect results. Power scene data involve privacy and security issues, and there is an imbalance in the number of samples across different defect categories, all of which will affect the performance of defect detection models. With the emergence of the Internet of Things (IoT), the integration of IoT with machine learning offers a new direction for defect detection in power equipment. Meanwhile, a generative adversarial network based on multi-view fusion and self-attention is proposed for few-shot image generation, named MVSA-GAN. The IoT devices capture real-time data from the power scene, which are then used to train the MVSA-GAN model, enabling it to generate realistic and diverse defect data. The designed self-attention encoder focuses on the relevant features of different parts of the image to capture the contextual information of the input image and improve the authenticity and coherence of the image. A multi-view feature fusion module is proposed to capture the complex structure and texture of the power scene through the selective fusion of global and local features, and improve the authenticity and diversity of generated images. Experiments show that the few-shot image generation method proposed in this paper can generate real and diverse defect data for power scene defects. The proposed method achieved FID and LPIPS scores of 67.87 and 0.179, surpassing SOTA methods, such as FIGR and DAWSON. [ABSTRACT FROM AUTHOR]

Deep unrolling networks (DUNs) have emerged as a promising approach for solving compressed sensing (CS) problems due to their superior explainability, speed, and performance compared to classical deep network models. However, the CS performance in terms of efficiency and accuracy remains a principal challenge for approaching further improvements. In this paper, we propose a novel deep unrolling model, SALSA-Net, to solve the image CS problem. The network architecture of SALSA-Net is inspired by unrolling and truncating the split augmented Lagrangian shrinkage algorithm (SALSA) which is used to solve sparsity-induced CS reconstruction problems. SALSA-Net inherits the interpretability of the SALSA algorithm while incorporating the learning ability and fast reconstruction speed of deep neural networks. By converting the SALSA algorithm into a deep network structure, SALSA-Net consists of a gradient update module, a threshold denoising module, and an auxiliary update module. All parameters, including the shrinkage thresholds and gradient steps, are optimized through end-to-end learning and are subject to forward constraints to ensure faster convergence. Furthermore, we introduce learned sampling to replace traditional sampling methods so that the sampling matrix can better preserve the feature information of the original signal and improve sampling efficiency. Experimental results demonstrate that SALSA-Net achieves significant reconstruction performance compared to state-of-the-art methods while inheriting the advantages of explainable recovery and high speed from the DUNs paradigm. [ABSTRACT FROM AUTHOR]

Considering the high incidence of accidents at tunnel construction sites, using robots to replace humans in hazardous tasks can effectively safeguard their lives. However, most robots currently used in this field require manual control and lack autonomous obstacle avoidance capability. To address these issues, we propose a lightweight model based on an improved version of YOLOv5 for obstacle detection. Firstly, to enhance detection speed and reduce computational load, we modify the backbone network to the lightweight Shufflenet v2. Secondly, we introduce a coordinate attention mechanism to enhance the network's ability to learn feature representations. Subsequently, we replace the neck convolution block with GSConv to improve the model's efficiency. Finally, we modify the model's upsampling method to further enhance detection accuracy. Through comparative experiments on the model, the results demonstrate that our approach achieves an approximately 37% increase in detection speed with a minimal accuracy reduction of 1.5%. The frame rate has improved by about 54%, the parameter count has decreased by approximately 74%, and the model size has decreased by 2.5 MB. The experimental results indicate that our method can reduce hardware requirements for the model, striking a balance between detection speed and accuracy. [ABSTRACT FROM AUTHOR]

We consider a complex control problem: making a monopod accurately reach a target with a single jump. The monopod can jump in any direction at different elevations of the terrain. This is a paradigm for a much larger class of problems, which are extremely challenging and computationally expensive to solve using standard optimization-based techniques. Reinforcement learning (RL) is an interesting alternative, but an end-to-end approach in which the controller must learn everything from scratch can be non-trivial with a sparse-reward task like jumping. Our solution is to guide the learning process within an RL framework leveraging nature-inspired heuristic knowledge. This expedient brings widespread benefits, such as a drastic reduction of learning time, and the ability to learn and compensate for possible errors in the low-level execution of the motion. Our simulation results reveal a clear advantage of our solution against both optimization-based and end-to-end RL approaches. [ABSTRACT FROM AUTHOR]

To solve the problems of low accuracy, low real-time performance, poor robustness and others caused by the complex environment, this paper proposes a face mask recognition and standard wear detection algorithm based on the improved YOLO-v4. Firstly, an improved CSPDarkNet53 is introduced into the trunk feature extraction network, which reduces the computing cost of the network and improves the learning ability of the model. Secondly, the adaptive image scaling algorithm can reduce computation and redundancy effectively. Thirdly, the improved PANet structure is introduced so that the network has more semantic information in the feature layer. At last, a face mask detection data set is made according to the standard wearing of masks. Based on the object detection algorithm of deep learning, a variety of evaluation indexes are compared to evaluate the effectiveness of the model. The results of the comparations show that the mAP of face mask recognition can reach 98.3% and the frame rate is high at 54.57 FPS, which are more accurate compared with the exiting algorithm. [ABSTRACT FROM AUTHOR]

When performing multiple target detection, it is difficult to detect small and occluded targets in complex traffic scenes. To this end, an improved YOLOv4 detection method is proposed in this work. Firstly, the network structure of the original YOLOv4 is adjusted, and the 4× down-sampling feature map of the backbone network is introduced into the neck network of the YOLOv4 model to splice the feature map with 8× down-sampling to form a four-scale detection structure, which enhances the fusion of deep and shallow semantics information of the feature map to improve the detection accuracy of small targets. Then, the convolutional block attention module (CBAM) is added to the model neck network to enhance the learning ability for features in space and on channels. Lastly, the detection rate of the occluded target is improved by using the soft non-maximum suppression (Soft-NMS) algorithm based on the distance intersection over union (DIoU) to avoid deleting the bounding boxes. On the KITTI dataset, experimental evaluation is performed and the analysis results demonstrate that the proposed detection model can effectively improve the multiple target detection accuracy, and the mean average accuracy (mAP) of the improved YOLOv4 model reaches 81.23%, which is 3.18% higher than the original YOLOv4; and the computation speed of the proposed model reaches 47.32 FPS. Compared with existing popular detection models, the proposed model produces higher detection accuracy and computation speed. [ABSTRACT FROM AUTHOR]

As a detection method, X-ray Computed Tomography (CT) technology has the advantages of clear imaging, short detection time, and low detection cost. This makes it more widely used in clinical disease screening, detection, and disease tracking. This study exploits the ability of sparse representation to learn sparse transformations of information and combines it with image decomposition theory. The structural information of low-dose CT images is separated from noise and artifact information, and the sparse expression of sparse transformation is used to improve the imaging effect. In this paper, two different learned sparse transformations are used. The first covers more organizational information about the scanned object. The other can cover more noise artifacts. Both methods can improve the ability to learn sparse transformations to express various image information. Experimental results show that the algorithm is effective. [ABSTRACT FROM AUTHOR]

In this study, we demonstrate a single-track magnetic code tape-based absolute position sensor system. Unlike traditional dual-track systems, our method simplifies manufacturing and avoids crosstalk between tracks, offering higher tolerance to alignment errors. The sensing system employs an array of magnetic field sensing elements that recognize the bit sequence encoded on the tape. This approach allows for accurate position determination even when the number of sensing elements is fewer than the number of bits covered, and without the need for specific spacing between sensing elements and bit length. We demonstrate the system's ability to learn and adapt to various magnetic code patterns, including those that are irregular or have been altered. Our method can identify and localize the sensed magnetic field pattern directly within a self-learned magnetic field map, providing robust performance in diverse conditions. This self-adaptive capability enhances operational safety and reliability, as the system can continue functioning even when the magnetic tape is misaligned or has undergone changes. [ABSTRACT FROM AUTHOR]

X-ray images typically contain complex background information and abundant small objects, posing significant challenges for object detection in security tasks. Most existing object detection methods rely on complex networks and high computational costs, which poses a challenge to implement lightweight models. This article proposes Fine-YOLO to achieve rapid and accurate detection in the security domain. First, a low-parameter feature aggregation (LPFA) structure is designed for the backbone feature network of YOLOv7 to enhance its ability to learn more information with a lighter structure. Second, a high-density feature aggregation (HDFA) structure is proposed to solve the problem of loss of local details and deep location information caused by the necked feature fusion network in YOLOv7-Tiny-SiLU, connecting cross-level features through max-pooling. Third, the Normalized Wasserstein Distance (NWD) method is employed to alleviate the convergence complexity resulting from the extreme sensitivity of bounding box regression to small objects. The proposed Fine-YOLO model is evaluated on the EDS dataset, achieving a detection accuracy of 58.3% with only 16.1 M parameters. In addition, an auxiliary validation is performed on the NEU-DET dataset, the detection accuracy reaches 73.1%. Experimental results show that Fine-YOLO is not only suitable for security, but can also be extended to other inspection areas. [ABSTRACT FROM AUTHOR]

As a non-destructive, fast, and cost-effective technique, near-infrared (NIR) spectroscopy has been widely used to determine the content of bioactive components in tea. However, due to the similar chemical structures of various catechins in black tea, the NIR spectra of black tea severely overlap in certain bands, causing nonlinear relationships and reducing analytical accuracy. In addition, the number of NIR spectral wavelengths is much larger than that of the modeled samples, and the small-sample learning problem is rather typical. These issues make the use of NIRS to simultaneously determine black tea catechins challenging. To address the above problems, this study innovatively proposed a wavelength selection algorithm based on feature interval combination sensitivity segmentation (FIC-SS). This algorithm extracts wavelengths at both coarse-grained and fine-grained levels, achieving higher accuracy and stability in feature wavelength extraction. On this basis, the study built four simultaneous prediction models for catechins based on extreme learning machines (ELMs), utilizing their powerful nonlinear learning ability and simple model structure to achieve simultaneous and accurate prediction of catechins. The experimental results showed that for the full spectrum, the ELM model has better prediction performance than the partial least squares model for epicatechin (EC), epicatechin gallate (ECG), epigallocatechin (EGC), and epigallocatechin gallate (EGCG). For the feature wavelengths, our proposed FIC-SS-ELM model enjoys higher prediction performance than ELM models based on other wavelength selection algorithms; it can simultaneously and accurately predict the content of EC ( R p 2 = 0.91, RMSEP = 0.019), ECG ( R p 2 = 0.96, RMSEP = 0.11), EGC ( R p 2 = 0.97, RMSEP = 0.15), and EGCG ( R p 2 = 0.97, RMSEP = 0.35) in black tea. The results of this study provide a new method for the quantitative determination of the bioactive components of black tea. [ABSTRACT FROM AUTHOR]

Wheat head detection can estimate various wheat traits, such as density, health, and the presence of wheat head. However, traditional detection methods have a huge array of problems, including low efficiency, strong subjectivity, and poor accuracy. In this paper, a method of wheat-head detection based on a deep neural network is proposed to enhance the speed and accuracy of detection. The YOLOv4 is taken as the basic network. The backbone part in the basic network is enhanced by adding dual spatial pyramid pooling (SPP) networks to improve the ability of feature learning and increase the receptive field of the convolutional network. Multilevel features are obtained by a multipath neck part using a top-down to bottom-up strategy. Finally, YOLOv3′s head structures are used to predict the boxes of wheat heads. For training images, some data augmentation technologies are used. The experimental results demonstrate that the proposed method has a significant advantage in accuracy and speed. The mean average precision of our method is 94.5%, and the detection speed is 71 FPS that can achieve the effect of real-time detection. [ABSTRACT FROM AUTHOR]

The identification of the growth and development period of rice is of great significance to achieve high-yield and high-quality rice. However, the acquisition of rice growth period information mainly relies on manual observation, which has problems such as low efficiency and strong subjectivity. In order to solve these problems, a lightweight recognition method is proposed to automatically identify the growth period of rice: Small-YOLOv5, which is based on improved YOLOv5s. Firstly, the new backbone feature extraction network MobileNetV3 was used to replace the YOLOv5s backbone network to reduce the model size and the number of model parameters, thus improving the detection speed of the model. Secondly, in the feature fusion stage of YOLOv5s, we introduced a more lightweight convolution method, GsConv, to replace the standard convolution. The computational cost of GsConv is about 60–70% of the standard convolution, but its contribution to the model learning ability is no less than that of the standard convolution. Based on GsConv, we built a lightweight neck network to reduce the complexity of the network model while maintaining accuracy. To verify the performance of Small-YOLOv5s, we tested it on a self-built dataset of rice growth period. The results show that compared with YOLOv5s (5.0) on the self-built dataset, the number of the model parameter was reduced by 82.4%, GFLOPS decreased by 85.9%, and the volume reduced by 86.0%. The mAP (0.5) value of the improved model was 98.7%, only 0.8% lower than that of the original YOLOv5s model. Compared with the mainstream lightweight model YOLOV5s- MobileNetV3-Small, the number of the model parameter was decreased by 10.0%, the volume reduced by 9.6%, and the mAP (0.5:0.95) improved by 5.0%—reaching 94.7%—and the recall rate improved by 1.5%—reaching 98.9%. Based on experimental comparisons, the effectiveness and superiority of the model have been verified. [ABSTRACT FROM AUTHOR]

In the field of surface defect detection, the scale difference of product surface defects is often huge. The existing defect detection methods based on Convolutional Neural Networks (CNNs) are more inclined to express macro and abstract features, and the ability to express local and small defects is insufficient, resulting in an imbalance of feature expression capabilities. In this paper, a Multi-Scale Feature Learning Network (MSF-Net) based on Dual Module Feature (DMF) extractor is proposed. DMF extractor is mainly composed of optimized Concatenated Rectified Linear Units (CReLUs) and optimized Inception feature extraction modules, which increases the diversity of feature receptive fields while reducing the amount of calculation; the feature maps of the middle layer with different sizes of receptive fields are merged to increase the richness of the receptive fields of the last layer of feature maps; the residual shortcut connections, batch normalization layer and average pooling layer are used to replace the fully connected layer to improve training efficiency, and make the multi-scale feature learning ability more balanced at the same time. Two representative multi-scale defect data sets are used for experiments, and the experimental results verify the advancement and effectiveness of the proposed MSF-Net in the detection of surface defects with multi-scale features. [ABSTRACT FROM AUTHOR]

At present, convolutional neural networks (CNNs) have been widely applied to the task of skin disease image segmentation due to the fact of their powerful information discrimination abilities and have achieved good results. However, it is difficult for CNNs to capture the connection between long-range contexts when extracting deep semantic features of lesion images, and the resulting semantic gap leads to the problem of segmentation blur in skin lesion image segmentation. In order to solve the above problems, we designed a hybrid encoder network based on transformer and fully connected neural network (MLP) architecture, and we call this approach HMT-Net. In the HMT-Net network, we use the attention mechanism of the CTrans module to learn the global relevance of the feature map to improve the network's ability to understand the overall foreground information of the lesion. On the other hand, we use the TokMLP module to effectively enhance the network's ability to learn the boundary features of lesion images. In the TokMLP module, the tokenized MLP axial displacement operation strengthens the connection between pixels to facilitate the extraction of local feature information by our network. In order to verify the superiority of our network in segmentation tasks, we conducted extensive experiments on the proposed HMT-Net network and several newly proposed Transformer and MLP networks on three public datasets (ISIC2018, ISBI2017, and ISBI2016) and obtained the following results. Our method achieves 82.39%, 75.53%, and 83.98% on the Dice index and 89.35%, 84.93%, and 91.33% on the IOU. Compared with the latest skin disease segmentation network, FAC-Net, our method improves the Dice index by 1.99%, 1.68%, and 1.6%, respectively. In addition, the IOU indicators have increased by 0.45%, 2.36%, and 1.13%, respectively. The experimental results show that our designed HMT-Net achieves state-of-the-art performance superior to other segmentation methods. [ABSTRACT FROM AUTHOR]

Multimodal sentiment analysis has gained popularity as a research field for its ability to predict users' emotional tendencies more comprehensively. The data fusion module is a critical component of multimodal sentiment analysis, as it allows for integrating information from multiple modalities. However, it is challenging to combine modalities and remove redundant information effectively. In our research, we address these challenges by proposing a multimodal sentiment analysis model based on supervised contrastive learning, which leads to more effective data representation and richer multimodal features. Specifically, we introduce the MLFC module, which utilizes a convolutional neural network (CNN) and Transformer to solve the redundancy problem of each modal feature and reduce irrelevant information. Moreover, our model employs supervised contrastive learning to enhance its ability to learn standard sentiment features from data. We evaluate our model on three widely-used datasets, namely MVSA-single, MVSA-multiple, and HFM, demonstrating that our model outperforms the state-of-the-art model. Finally, we conduct ablation experiments to validate the efficacy of our proposed method. [ABSTRACT FROM AUTHOR]

The application of transfer learning in fault diagnosis has been developed in recent years. It can use existing data to solve the problem of fault recognition under different working conditions. Due to the complexity of the equipment and the openness of the working environment in industrial production, the status of the equipment is changeable, and the collected signals can have new fault classes. Therefore, the open set recognition ability of the transfer learning method is an urgent research direction. The existing transfer learning model can have a severe negative transfer problem when solving the open set problem, resulting in the aliasing of samples in the feature space and the inability to separate the unknown classes. To solve this problem, we propose a Weighted Domain Adaptation with Double Classifiers (WDADC) method. Specifically, WDADC designs the weighting module based on Jensen–Shannon divergence, which can evaluate the similarity between each sample in the target domain and each class in the source domain. Based on this similarity, a weighted loss is constructed to promote the positive transfer between shared classes in the two domains to realize the recognition of shared classes and the separation of unknown classes. In addition, the structure of double classifiers in WDADC can mitigate the overfitting of the model by maximizing the discrepancy, which helps extract the domain-invariant and class-separable features of the samples when the discrepancy between the two domains is large. The model's performance is verified in several fault datasets of rotating machinery. The results show that the method is effective in open set fault diagnosis and superior to the common domain adaptation methods. [ABSTRACT FROM AUTHOR]

In this paper we present a mobile augmented reality (MAR) application supporting teaching activities in interior design. The application supports students in learning interior layout design, interior design symbols, and the effects of different design layout decisions. Utilizing the latest AR technology, users can place 3D models of virtual objects as e.g., chairs or tables on top of a design layout plan and interact with these on their mobile devices. Students can experience alternative design decision in real-time and increases the special perception of interior designs. Our system fully supports the import of interior deployment layouts and the generation of 3D models based on design artefacts based on typical design layout plan design symbols and allows the user to investigate different design alternatives. We applied John Keller's Attention, Relevance, Confidence, and Satisfaction (ARCS) learning motivation model to validate our solution to examine the students' willingness and verify the ability of students to improve learning through MAR technology. We compared a sample experimental group of N = 52 test-subjects with a sample of N = 48 candidates in a control group. Learning indicators as learning interest, confidence, satisfaction and effective have been utilized to assess the students' learning motivation through the use of MAR technology. The learning results have been determined by the independent sample t testing. The significance of the post-test had a p-value < 0.05 difference. The result of the study clearly shows that the reference group utilizing MAR technology as a learning aid show a higher learning effectiveness as the control group. Thus, we conclude that MAR technology does enhance students' learning ability for interior design and making appropriate design decisions. [ABSTRACT FROM AUTHOR]

Contrastive learning has received increasing attention in the field of skeleton-based action representations in recent years. Most contrastive learning methods use simple augmentation strategies to construct pairs of positive samples. When using such pairs of positive samples to learn action representations, deeper feature information cannot be learned, thus affecting the performance of downstream tasks. To solve the problem of insufficient learning ability, we propose an asymmetric data augmentation strategy and attempt to apply it to the training of 3D skeleton-based action representations. First, we carefully study the different characteristics presented by different skeleton views and choose a specific augmentation method for a certain view. Second, specific augmentation methods are incorporated into the left and right branches of the asymmetric data augmentation pipeline to increase the convergence difficulty of the contrastive learning task, thereby significantly improving the quality of the learned action representations. Finally, since many methods directly act on the joint view, the augmented samples are quite different from the original samples. We use random probability activation to transform the joint view to avoid extreme augmentation of the joint view. Extensive experiments on NTU RGB + D datasets show that our method is effective. [ABSTRACT FROM AUTHOR]

This paper proposes a novel algorithm for image phase retrieval, i.e., for recovering complex-valued images from the amplitudes of noisy linear combinations (often the Fourier transform) of the sought complex images. The algorithm is developed using the alternating projection framework and is aimed to obtain high performance for heavily noisy (Poissonian or Gaussian) observations. The estimation of the target images is reformulated as a sparse regression, often termed sparse coding, in the complex domain. This is accomplished by learning a complex domain dictionary from the data it represents via matrix factorization with sparsity constraints on the code (i.e., the regression coefficients). Our algorithm, termed dictionary learning phase retrieval (DLPR), jointly learns the referred to dictionary and reconstructs the unknown target image. The effectiveness of DLPR is illustrated through experiments conducted on complex images, simulated and real, where it shows noticeable advantages over the state-of-the-art competitors. [ABSTRACT FROM AUTHOR]

Fluorescence lifetime imaging (FLIM) is a powerful tool that provides unique quantitative information for biomedical research. In this study, we propose a multi-layer-perceptron-based mixer (MLP-Mixer) deep learning (DL) algorithm named FLIM-MLP-Mixer for fast and robust FLIM analysis. The FLIM-MLP-Mixer has a simple network architecture yet a powerful learning ability from data. Compared with the traditional fitting and previously reported DL methods, the FLIM-MLP-Mixer shows superior performance in terms of accuracy and calculation speed, which has been validated using both synthetic and experimental data. All results indicate that our proposed method is well suited for accurately estimating lifetime parameters from measured fluorescence histograms, and it has great potential in various real-time FLIM applications. [ABSTRACT FROM AUTHOR]

The growth of the Internet has expanded the amount of data expressed by users across multiple platforms. The availability of these different worldviews and individuals' emotions empowers sentiment analysis. However, sentiment analysis becomes even more challenging due to a scarcity of standardized labeled data in the Bangla NLP domain. The majority of the existing Bangla research has relied on models of deep learning that significantly focus on context-independent word embeddings, such as Word2Vec, GloVe, and fastText, in which each word has a fixed representation irrespective of its context. Meanwhile, context-based pre-trained language models such as BERT have recently revolutionized the state of natural language processing. In this work, we utilized BERT's transfer learning ability to a deep integrated model CNN-BiLSTM for enhanced performance of decision-making in sentiment analysis. In addition, we also introduced the ability of transfer learning to classical machine learning algorithms for the performance comparison of CNN-BiLSTM. Additionally, we explore various word embedding techniques, such as Word2Vec, GloVe, and fastText, and compare their performance to the BERT transfer learning strategy. As a result, we have shown a state-of-the-art binary classification performance for Bangla sentiment analysis that significantly outperforms all embedding and algorithms. [ABSTRACT FROM AUTHOR]

Few-shot learning (FSL) is of great significance to the field of machine learning. The ability to learn and generalize using a small number of samples is an obvious distinction between artificial intelligence and humans. In the FSL domain, most graph neural networks (GNNs) focus on transferring labeled sample information to an unlabeled query sample, ignoring the important role of semantic information during the classification process. Our proposed method embeds semantic information of classes into a GNN, creating a word embedding distribution propagation graph network (WPGN) for FSL. We merge the attention mechanism with our backbone network, use the Mahalanobis distance to calculate the similarity of classes, select the Funnel ReLU (FReLU) function as the activation function of the Transform layer, and update the point graph and word embedding distribution graph. In extensive experiments on FSL benchmarks, compared with the baseline model, the accuracy of the WPGN on the 5-way-1/2/5 shot tasks increased by 9.03, 4.56, and 4.15%, respectively. [ABSTRACT FROM AUTHOR]

High precision Global Navigation Satellite System (GNSS) measurements are becoming more and more popular in alpine skiing due to the relatively undemanding setup and excellent performance. However, GNSS provides only single-point measurements that are defined with the antenna placed typically behind the skier's neck. A key issue is how to estimate other more relevant parameters of the skier's body, like the center of mass (COM) and ski trajectories. Previously, these parameters were estimated by modeling the skier's body with an inverted-pendulum model that oversimplified the skier's body. In this study, we propose two machine learning methods that overcome this shortcoming and estimate COM and skis trajectories based on a more faithful approximation of the skier's body with nine degrees-of-freedom. The first method utilizes a well-established approach of artificial neural networks, while the second method is based on a state-of-the-art statistical generalization method. Both methods were evaluated using the reference measurements obtained on a typical giant slalom course and compared with the inverted-pendulum method. Our results outperform the results of commonly used inverted-pendulum methods and demonstrate the applicability of machine learning techniques in biomechanical measurements of alpine skiing. [ABSTRACT FROM AUTHOR]

Signal features can be obscured in noisy environments, resulting in low accuracy of radar emitter signal recognition based on traditional methods. To improve the ability of learning features from noisy signals, a new radar emitter signal recognition method based on one-dimensional (1D) deep residual shrinkage network (DRSN) is proposed, which offers the following advantages: (i) Unimportant features are eliminated using the soft thresholding function, and the thresholds are automatically set based on the attention mechanism; (ii) without any professional knowledge of signal processing or dimension conversion of data, the 1D DRSN can automatically learn the features characterizing the signal directly from the 1D data and achieve a high recognition rate for noisy signals. The effectiveness of the 1D DRSN was experimentally verified under different types of noise. In addition, comparison with other deep learning methods revealed the superior performance of the DRSN. Last, the mechanism of eliminating redundant features using the soft thresholding function was analyzed. [ABSTRACT FROM AUTHOR]

Determining ingestive behaviors of dairy cows is critical to evaluate their productivity and health status. The objectives of this research were to (1) develop the relationship between forage species/heights and sound characteristics of three different ingestive behaviors (bites, chews, and chew-bites); (2) comparatively evaluate three deep learning models and optimization strategies for classifying the three behaviors; and (3) examine the ability of deep learning modeling for classifying the three ingestive behaviors under various forage characteristics. The results show that the amplitude and duration of the bite, chew, and chew-bite sounds were mostly larger for tall forages (tall fescue and alfalfa) compared to their counterparts. The long short-term memory network using a filtered dataset with balanced duration and imbalanced audio files offered better performance than its counterparts. The best classification performance was over 0.93, and the best and poorest performance difference was 0.4–0.5 under different forage species and heights. In conclusion, the deep learning technique could classify the dairy cow ingestive behaviors but was unable to differentiate between them under some forage characteristics using acoustic signals. Thus, while the developed tool is useful to support precision dairy cow management, it requires further improvement. [ABSTRACT FROM AUTHOR]

At present, the one-stage detector based on the lightweight model can achieve real-time speed, but the detection performance is challenging. To enhance the discriminability and robustness of the model extraction features and improve the detector's detection performance for small objects, we propose two modules in this work. First, we propose a receptive field enhancement method, referred to as adaptive receptive field fusion (ARFF). It enhances the model's feature representation ability by adaptively learning the fusion weights of different receptive field branches in the receptive field module. Then, we propose an enhanced up-sampling (EU) module to reduce the information loss caused by up-sampling on the feature map. Finally, we assemble ARFF and EU modules on top of YOLO v3 to build a real-time, high-precision and lightweight object detection system referred to as the ARFF-EU network. We achieve a state-of-the-art speed and accuracy trade-off on both the Pascal VOC and MS COCO data sets, reporting 83.6% AP at 37.5 FPS and 42.5% AP at 33.7 FPS, respectively. The experimental results show that our proposed ARFF and EU modules improve the detection performance of the ARFF-EU network and achieve the development of advanced, very deep detectors while maintaining real-time speed. [ABSTRACT FROM AUTHOR]

Oxford Nanopore sequencing is an important sequencing technology, which reads the nucleotide sequence by detecting the electrical current signal changes when DNA molecule is forced to pass through a biological nanopore. The research on signal simulation of nanopore sequencing is highly desirable for method developments of nanopore sequencing applications. To improve the simulation accuracy, we propose a novel signal simulation method based on Bi-directional Gated Recurrent Units (BiGRU). In this method, the signal processing model based on BiGRU is built to replace the traditional low-pass filter to post-process the ground-truth signal calculated by the input nucleotide sequence and nanopore sequencing pore model. Gaussian noise is then added to the filtered signal to generate the final simulated signal. This method can accurately model the relation between ground-truth signal and real-world sequencing signal through experimental sequencing data. The simulation results reveal that the proposed method utilizing the powerful learning ability of the neural network can generate the simulated signal that is closer to the real-world sequencing signal in the time and frequency domains than the existing simulation method. [ABSTRACT FROM AUTHOR]