Search

In recent years, the ability to accurately predict age and gender from facial images has gained significant traction across various fields such as personalized marketing, human–computer interaction, and security surveillance. However, the high computational cost of the current models limits their practicality for real-time applications on resource-constrained devices. This study addressed this challenge by leveraging knowledge distillation to develop lightweight age and gender prediction models that maintain a high accuracy. We propose a knowledge distillation method using teacher bounds for the efficient learning of small models for age and gender. This method allows the student model to selectively receive the teacher model's knowledge, preventing it from unconditionally learning from the teacher in challenging age/gender prediction tasks involving factors like illusions and makeup. Our experiments used MobileNetV3 and EfficientFormer as the student models and Vision Outlooker (VOLO)-D1 as the teacher model, resulting in substantial efficiency improvements. MobileNetV3-Small, one of the student models we experimented with, achieved a 94.27% reduction in parameters and a 99.17% reduction in Giga Floating Point Operations per Second (GFLOPs). Furthermore, the distilled MobileNetV3-Small model improved gender prediction accuracy from 88.11% to 90.78%. Our findings confirm that knowledge distillation can effectively enhance model performance across diverse demographic groups while ensuring efficiency for deployment on embedded devices. This research advances the development of practical, high-performance AI applications in resource-limited environments. [ABSTRACT FROM AUTHOR]

Potato crop has become integral part of our diet due to its wide use in variety of dishes, making it an important food crop. Its importance also stems from the fact that it is one of the cheapest vegetables available throughout the year. This makes it crucial to keep potato prices affordable for developing countries where the majority of the population falls under the middle-income bracket. Consequently, there is a need to develop a robust, effective, and portable technique to detect diseases in potato plant leaves. In this work, an attention-based disease detection technique is proposed. This technique selectively focuses on specific parts of an image which reveal the disease. This technique leverages transfer learning combined with two attention modules: the channel attention module and spatial attention module. By focusing on specific parts of the images, the proposed technique is able to achieve almost similar accuracy with significantly fewer parameters. The proposed technique has been validated using four pre-trained models: DenseNet169, XceptionNet, MobileNet, and VGG16. All of these models are able to achieve almost the same level of training and validation accuracy, around 90–97%, even after reducing the number of parameters by 40–50%. It shows that the proposed technique effectively reduces model complexity without compromising performance. [ABSTRACT FROM AUTHOR]

The daunting issues that are promptly faced worldwide are the sophisticated cyber-attacks in all kinds of organizations and applications. The development of cloud computing pushed organizations to shift their business towards the virtual machines of the cloud. Nonetheless, the lack of security throughout the programmatic and declarative levels explicitly prone to cyber-attacks in the cloud platform. The exploitation of web pages and the cloud is due to the uncrated open web application security projects (OWASP) fragilities and fragilities in the cloud containers and network resources. With the utilization of advanced hacking vectors, the attackers attack data integrity, confidentiality, and availability. Hence, it’s ineluctable to frame the application security-based technique for the reduction of attacks. In concern to this, we propose a novel Deep learning-based secured advanced web application firewall to overcome the lack of missing programmatic and declarative level securities in the application. For this, we adopted the MobileNet-based technique to ensure the assurance of security. Simulations are effectuated and analyzed the robustness with the statistical parameters such as accuracy, precision, sensitivity, and specificity and made the comparative study with the existing works. Our proposed technique surpasses all the other techniques and provides better security in the cloud. [ABSTRACT FROM AUTHOR]

Heart muscle damage is a result of myocardial infarction (MI), which is caused by inadequate blood supply. Around the world, MI is the leading cause of death for middle-aged and older people. To reduce the risk of MI, early detection is important. This detection is obtained by using a deep learning algorithm. In the literature, few methods are reviewed which does not provide optimal results for detection. Hence, in this paper, the Enhanced AlexNet is developed for an effective diagnosis (ED) to identify MI signals (EAlexNet). To train AlexNet and obtain the best results, a hybrid spider monkey optimization (SMO) and salp swarm optimization (SSO) algorithm is used. Four phases are taken into consideration in the paper to find the MI signals. The input dataset is used to construct the echo frames, and the formed frames are then trained using the EAlexNet. Then, using an adaptive algorithm called a support vector machine (SVM) with kernel function, the process of feature extraction is carried out. Finally, the proposed approach is used to complete the MI classification process. The normal (non-MI) and abnormal (MI) cases are identified from the proposed model. The HMC-QU dataset is taken into account for analysis purposes, and the effectiveness of the suggested strategy is assessed. The suggested approach is contrasted with the current approaches, including ResNet, MobileNet, and VGGNet, respectively. The suggested method is put into practise using the MATLAB platform, and the accuracy, sensitivity, precision, and specificity performance analysis is examined. [ABSTRACT FROM AUTHOR]

With the rapid development of the Industrial Internet of Things in rotating machinery, the amount of data sampled by mechanical vibration wireless sensor networks (MvWSNs) has increased significantly, straining bandwidth capacity. Concurrently, the safety requirements for rotating machinery have escalated, necessitating enhanced real-time data processing capabilities. Conventional methods, reliant on experiential approaches, have proven inefficient in meeting these evolving challenges. To this end, a fault detection method for rotating machinery based on mobileNet in MvWSNs is proposed to address these intractable issues. The small and light deep learning model is helpful to realize nearly real-time sensing and fault detection, lightening the communication pressure of MvWSNs. The well-trained deep learning is implanted on the MvWSNs sensor node, an edge computing platform developed via embedded STM32 microcontrollers (STMicroelectronics International NV, Geneva, Switzerland). Data acquisition, data processing, and data classification are all executed on the computing- and energy-constrained sensor node. The experimental results demonstrate that the proposed fault detection method can achieve about 0.99 for the DDS dataset and an accuracy of 0.98 in the MvWSNs sensor node. Furthermore, the final transmission data size is only 0.1% compared to the original data size. It is also a time-saving method that can be accomplished within 135 ms while the raw data will take about 1000 ms to transmit to the monitoring center when there are four sensor nodes in the network. Thus, the proposed edge computing method shows good application prospects in fault detection and control of rotating machinery with high time sensitivity. [ABSTRACT FROM AUTHOR]

Considering factors such as illumination, camera quality variations, and background-specific variations, identifying a face using a smartphone-based facial image capture application is challenging. Face Image Quality Assessment refers to the process of taking a face image as input and producing some form of "quality" estimate as an output. Typically, quality assessment techniques use deep learning methods to categorize images. The models used in deep learning are shown as black boxes. This raises the question of the trustworthiness of the models. Several explainability techniques have gained importance in building this trust. Explainability techniques provide visual evidence of the active regions within an image on which the deep learning model makes a prediction. Here, we developed a technique for reliable prediction of facial images before medical analysis and security operations. A combination of gradient-weighted class activation mapping and local interpretable model-agnostic explanations were used to explain the model. This approach has been implemented in the preselection of facial images for skin feature extraction, which is important in critical medical science applications. We demonstrate that the use of combined explanations provides better visual explanations for the model, where both the saliency map and perturbation-based explainability techniques verify predictions. [ABSTRACT FROM AUTHOR]

Rail surface defects typically serve as early indicators of railway malfunctions, which may compromise the quality and corrosion resistance of rails, thereby endangering the safe operation of trains. The timely detection of defects is essential to ensure the safe operation of railways. To improve the classification accuracy of rail surface defect detection, this paper proposes a rail surface defects detection algorithm based on MobileNet-YOLOv7. By integrating lightweight deep learning algorithms into the engineering application of rail surface defect detection, a MobileNetV3 lightweight network is used as the backbone network for YOLOv7 to enhance both speed and accuracy in complex defect extraction. Subsequently, the efficient intersection over union (EIOU) loss function is utilized as the positional loss function to bolster system resilience. Finally, the k-means++ clustering algorithm is applied to obtain new anchor boxes. The experimental results demonstrate the effectiveness of the proposed method, achieving superior detection accuracy compared with traditional algorithms. [ABSTRACT FROM AUTHOR]

Texture recognition is a pivotal task in computer vision, crucial for applications in material sciences, medicine, and agriculture. Leveraging advancements in Deep Neural Networks (DNNs), researchers seek robust methods to discern intricate patterns in images. In the context of the burgeoning Tactile Internet (TI), efficient texture recognition algorithms are essential for real-time applications. This paper introduces a method named Local Binary Convolution Network with Intra-class Normalization (LBCNIN) for texture recognition. Incorporating features from the last layer of the backbone, LBCNIN employs a non-trainable Local Binary Convolution (LBC) layer, inspired by Local Binary Patterns (LBP), without fine-tuning the backbone. The encoded feature vector is fed into a linear Support Vector Machine (SVM) for classification, serving as the only trainable component. In the context of TI, the availability of images from multiple views, such as in 3D object semantic segmentation, allows for more data per object. Consequently, LBCNIN processes batches where each batch contains images from the same material class, with batch normalization employed as an intra-class normalization method, aiming to produce better results than single images. Comprehensive evaluations across texture benchmarks demonstrate LBCNIN's ability to achieve very good results under different resource constraints, attributed to the variability in backbone architectures. [ABSTRACT FROM AUTHOR]

To address the current inefficiencies and subjective nature of manual observation in maize cultivation, with the aim of achieving high efficiency and productivity, this study focused on the DeMaya D3 maize variety. It proposes a maize growth stage recognition method based on the MobileNet model, which is a lightweight convolutional neural network architecture. The method was tested and achieved recognition accuracies of 0.98, 0.96, 0.92, 0.85, and 0.97 for different growth stages, respectively. Additionally, a maize growth prediction model was developed. Based on data collected from experimental plots regarding maize plant height and stem diameter, the Prophet model and an optimized version of the Prophet model were used to forecast maize growth trends. The Prophet model is an open-source tool for time series forecasting. Comparative analysis was conducted between the predictions of the original Prophet model and the optimized version. The relative errors of the Prophet model predictions were 0.85%, 2.11%, and 0.79%, while those of the optimized Prophet model were 0.76%, 0.47%, and 0.71%. Compared to the Prophet model, the optimized model reduced errors by 0.09%, 1.64%, and 0.08%, respectively. The maize plant growth control system was designed to obtain the information through the collection layer. The decision-making layer judged the soil nutrient absorption and growth status. Finally, the management layer controlled water and fertilizer. [ABSTRACT FROM AUTHOR]

Copyright of Baghdad Science Journal is the property of Republic of Iraq Ministry of Higher Education & Scientific Research (MOHESR) and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)

Emotion is a psychological and physiological response to an event or stimulus. Understanding students' emotions helps teachers and educators interact more effectively with students and create a better learning environment. The importance of understanding students' emotions in the learning process has led to exploring the use of facial emotion classification technology. In this research, an ensemble approach consisting of ResNet, MobileNet, and Inception is applied to identify emotional expressions on the faces of school students using a dataset that includes emotions such as happiness, sadness, anger, surprise, and boredom, acquired from students of Darul Imarah State Junior High School, Great Aceh District, Indonesia. Our dataset is available publicly, and so-called USK-FEMO. The performance evaluation results show that each model and approach has significant capabilities in classifying facial emotions. The ResNet model shows the best performance with the highest accuracy, precision, recall, and F1-score, which is 86%. MobileNet and Inception also demonstrate good performance, indicating potential in handling complex expression variations. The most interesting finding is that the ensemble approach achieves the highest accuracy, precision, recall, and F1-score of 90%. By combining predictions from the three models, the ensemble approach can consistently and accurately address emotion variations. Implementing emotion classification models, individually and in an ensemble format, can improve teacher-student interactions and optimize learning strategies that are responsive to students' emotional needs.

Today, as in every life-threatening disease, early diagnosis of brain tumors plays a life-saving role. The brain tumor is formed by the transformation of brain cells from their normal structures into abnormal cell structures. These formed abnormal cells begin to form in masses in the brain regions. Nowadays, many different techniques are employed to detect these tumor masses, and the most common of these techniques is Magnetic Resonance Imaging (MRI). In this study, it is aimed to automatically detect brain tumors with the help of ensemble deep learning architectures (ResNet50, VGG19, InceptionV3 and MobileNet) and Class Activation Maps (CAMs) indicators by employing MRI images. The proposed system was implemented in three stages. In the first stage, it was determined whether there was a tumor in the MR images (Binary Approach). In the second stage, different tumor types (Normal, Glioma Tumor, Meningioma Tumor, Pituitary Tumor) were detected from MR images (Multi-class Approach). In the last stage, CAMs of each tumor group were created as an alternative tool to facilitate the work of specialists in tumor detection. The results showed that the overall accuracy of the binary approach was calculated as 100% on the ResNet50, InceptionV3 and MobileNet architectures, and 99.71% on the VGG19 architecture. Moreover, the accuracy values of 96.45% with ResNet50, 93.40% with VGG19, 85.03% with InceptionV3 and 89.34% with MobileNet architectures were obtained in the multi-class approach.

The inspection of packaging defects is a crucial aspect of maintaining the quality of industrial production, especially in the case of boxed products. This study introduces a novel approach for detecting physical defects in product packaging boxes by integrating image processing with deep learning, specifically transfer learning with two images as an input. The proposed method utilizes both top view and side view images of the packaging to determine its condition, a significant departure from the conventional single image input. Our approach incorporates 16 pre-trained model variants from EfficientNetV2, MobileNetV3, and ResNetV2 for transfer learning as feature extractors. The experimental findings demonstrate that the best model that leverages EfficientNetV2 variant achieves 100% accuracy and F1 score in terms of classification performance. However, the most optimal model in terms of classification performance and inference speed was the one that leveraged ResNetV2 variant. This model scored 95% accuracy and 95.24% F1 score, with an inference speed of 91 ms per prediction. [ABSTRACT FROM AUTHOR]

Deaf and hard-of-hearing people mainly communicate using sign language, which is a set of signs made using hand gestures combined with facial expressions to make meaningful and complete sentences. The problem that faces deaf and hard-of-hearing people is the lack of automatic tools that translate sign languages into written or spoken text, which has led to a communication gap between them and their communities. Most state-of-the-art vision-based sign language recognition approaches focus on translating non-Arabic sign languages, with few targeting the Arabic Sign Language (ArSL) and even fewer targeting the Saudi Sign Language (SSL). This paper proposes a mobile application that helps deaf and hard-of-hearing people in Saudi Arabia to communicate efficiently with their communities. The prototype is an Android-based mobile application that applies deep learning techniques to translate isolated SSL to text and audio and includes unique features that are not available in other related applications targeting ArSL. The proposed approach, when evaluated on a comprehensive dataset, has demonstrated its effectiveness by outperforming several state-of-the-art approaches and producing results that are comparable to these approaches. Moreover, testing the prototype on several deaf and hard-of-hearing users, in addition to hearing users, proved its usefulness. In the future, we aim to improve the accuracy of the model and enrich the application with more features. [ABSTRACT FROM AUTHOR]

Today, as in every life-threatening disease, early diagnosis of brain tumors plays a life-saving role. The brain tumor is formed by the transformation of brain cells from their normal structures into abnormal cell structures. These formed abnormal cells begin to form in masses in the brain regions. Nowadays, many different techniques are employed to detect these tumor masses, and the most common of these techniques is Magnetic Resonance Imaging (MRI). In this study, it is aimed to automatically detect brain tumors with the help of ensemble deep learning architectures (ResNet50, VGG19, InceptionV3 and MobileNet) and Class Activation Maps (CAMs) indicators by employing MRI images. The proposed system was implemented in three stages. In the first stage, it was determined whether there was a tumor in the MR images (Binary Approach). In the second stage, different tumor types (Normal, Glioma Tumor, Meningioma Tumor, Pituitary Tumor) were detected from MR images (Multi-class Approach). In the last stage, CAMs of each tumor group were created as an alternative tool to facilitate the work of specialists in tumor detection. The results showed that the overall accuracy of the binary approach was calculated as 100% on the ResNet50, InceptionV3 and MobileNet architectures, and 99.71% on the VGG19 architecture. Moreover, the accuracy values of 96.45% with ResNet50, 93.40% with VGG19, 85.03% with InceptionV3 and 89.34% with MobileNet architectures were obtained in the multi-class approach. [ABSTRACT FROM AUTHOR]

Epilepsy is considered the most common cerebral disorder, around 1% of the worldwide population suffer from it. Recently, detection of epilepsy has attracted more and more attention. It has become a hastily increasing problem that can worsen their conditions which necessitate a specific and crucial attention where the symptoms can be an impaired awareness or motor symptoms. Besides that, the difficult process of manual inspection of electroencephalography electroencephalogram (EEG). This paper proposes using transfer learning models to detect both normal and epileptic brain activity and auto-classify signals from the brain. The models considered for this study are GoogleNet, MobileNet, and inception residual neural network inception ResNet. These models were associated with seven different classifiers such as discriminant. These classifiers were tested, analyzed and compared with each other. The efficiency of models is comparatively evaluated through result using multiple metrics. We therefore attained an accuracy of 96.53%, a precision of 97.18%, a false positive rate of 2.78% and an F1-score of 96.50%. Finally, comparison of the suggested approach with existing research shows that the performance of epilepsy classification has been markedly enhanced. [ABSTRACT FROM AUTHOR]

view of the large amount of garbage and the fact that a picture contains multiple garbage objects, this study proposes a garbage detection and classification algorithm based on the improved MobileNet network, which integrates the MobileNet network into YOLOv5(You Only Look Oncev5) target detection algorithm. At the same time, the CBAM(Convolutional Block Attention Modul) module is introduced in the backbone to filter meaningful information, and the vision transformer is used to aggregate and form image features. In addition, the weighted bidirectional feature pyramid network is used to distinguish the contribution of different features. At the same time, the ECA(Efficient Channel Attention) module is introduced to combine the image features and transmit them to the prediction layer. Finally, in order to obtain better performance when there is occlusion between garbage targets, soft-NMS(soft-Non Maximum Suppression) method and Alpha-IoU(Alpha-Intersection over Union) loss function is used to predict the extracted features. The experimental results show that the method proposed in this study can realize the location and recognition of multi-target and multi-category garbage., and the mAP(mean Average Percision) value reaches 90.31%, which is 4.95% higher than that of YOLOv5 network, and the processing speed is shortened by about 2.4 seconds. Compared with the Faster R-CNN(Region-based Convolutional Neural Network) algorithm which integrates ResNet(Residual Network) network, the algorithm proposed in this study improves the processing efficiency on the premise of ensuring the accuracy. [ABSTRACT FROM AUTHOR]

To overcome the challenges of inadequate representation and ineffective information exchange stemming from feature homogenization in underwater acoustic target recognition, we introduce a hybrid network named Mobile_ViT, which synergizes MobileNet and Transformer architectures. The network begins with a convolutional backbone incorporating an embedded coordinate attention mechanism to enhance the local details of inputs. This mechanism captures the long-term temporal dependencies and precise frequency–domain relationships of signals, focusing the features on the time–frequency positions. Subsequently, the Transformer's Encoder is integrated at the end of the backbone to facilitate global characterization, thus effectively overcoming the convolutional neural network's shortcomings in capturing long-range feature dependencies. Evaluation on the Shipsear and DeepShip datasets yields accuracies of 98.50% and 94.57%, respectively, marking a substantial improvement over the baseline. Notably, the proposed method also demonstrates obvious separation coefficients, signifying enhanced clustering effectiveness, and is lighter than other Transformers. [ABSTRACT FROM AUTHOR]

This study addresses the critical challenge of real-time identification of tomato leaf diseases using edge computing. Traditional plant disease detection methods rely on centralized cloud-based solutions that suffer from latency issues and require substantial bandwidth, making them less viable for real-time applications in remote or bandwidth-constrained environments. In response to these limitations, this study proposes an on-the-edge processing framework employing Convolutional Neural Networks (CNNs) to identify tomato diseases. This approach brings computation closer to the data source, reducing latency and conserving bandwidth. This study evaluates various pre-trained models, including MobileNetV2, InceptionV3, ResNet50, and VGG19 against a custom CNN, training and validating them on a comprehensive dataset of tomato leaf images. MobileNetV2 demonstrated exceptional performance, achieving an accuracy of 98.99%. The results highlight the potential of edge AI to revolutionize disease detection in agricultural settings, offering a scalable, efficient, and responsive solution that can be integrated into broader smart farming systems. This approach not only improves disease detection accuracy but can also provide actionable insights and timely alerts to farmers, ultimately contributing to increased crop yields and food security.

Timely and unbiased evaluation of Autism Spectrum Disorder (ASD) is essential for providing lasting benefits to affected individuals. However, conventional ASD assessment heavily relies on subjective criteria, lacking objectivity. Recent advancements propose the integration of modern processes, including artificial intelligence-based eye-tracking technology, for early ASD assessment. Nonetheless, the current diagnostic procedures for ASD often involve specialized investigations that are both time-consuming and costly, heavily reliant on the proficiency of specialists and employed techniques. To address the pressing need for prompt, efficient, and precise ASD diagnosis, an exploration of sophisticated intelligent techniques capable of automating disease categorization was presented. This study has utilized a freely accessible dataset comprising 547 eye-tracking systems that can be used to scan pathways obtained from 328 characteristically emerging children and 219 children with autism. To counter overfitting, state-of-the-art image resampling approaches to expand the training dataset were employed. Leveraging deep learning algorithms, specifically MobileNet, VGG19, DenseNet169, and a hybrid of MobileNet-VGG19, automated classifiers, that hold promise for enhancing diagnostic precision and effectiveness, was developed. The MobileNet model demonstrated superior performance compared to existing systems, achieving an impressive accuracy of 100%, while the VGG19 model achieved 92% accuracy. These findings demonstrate the potential of eye-tracking data to aid physicians in efficiently and accurately screening for autism. Moreover, the reported results suggest that deep learning approaches outperform existing event detection algorithms, achieving a similar level of accuracy as manual coding. Users and healthcare professionals can utilize these classifiers to enhance the accuracy rate of ASD diagnosis. The development of these automated classifiers based on deep learning algorithms holds promise for enhancing the diagnostic precision and effectiveness of ASD assessment, addressing the pressing need for prompt, efficient, and precise ASD diagnosis.

Pick-and-place operations are an integral part of robotic automation and smart manufacturing. By utilizing deep learning techniques on resource-constraint embedded devices, the pick-and-place operations can be made more accurate, efficient, and sustainable, compared to the high-powered computer solution. In this study, we propose a new technique for object detection on an embedded system using SSD Mobilenet V2 FPN Lite with the optimisation of the hyperparameter and image enhancement. By increasing the Red Green Blue (RGB) saturation level of the images, we gain a 7% increase in mean Average Precision (mAP) when compared to the control group and a 20% increase in mAP when compared to the COCO 2017 validation dataset. Using a Learning Rate of 0.08 with an Edge Tensor Processing Unit (TPU), we obtain high real-time detection scores of 97%. The high detection scores are important to the control algorithm, which uses the bounding box to send a signal to the collaborative robot for pick-and-place operation.

The COVID-19 pandemic has underscored the importance of face masks in curbing viral transmission, prompting governments worldwide to enforce stringent public health mandates requiring mask usage in public areas. Consequently, there is a growing focus on developing automated mask detection technologies to augment these measures and minimize viral spread. In this study, we explore the potential of the Swin Transformer architecture for accurately identifying face mask usage, aiming to surpass the current performance limitations of existing face mask detection models. We evaluate the performance of our proposed model and comparison models using comprehensive evaluation metrics, including accuracy, precision, recall, specificity, F1-score, Kappa coefficient, and MCC. Our experiments yield several notable findings. Firstly, MobileNetV2 demonstrates superior performance compared to the baseline CNN model across all seven evaluation metrics within the face mask datasets. Secondly, within the category of convolutional neural networks (CNNs), EfficientNetV2 outperforms MobileNetV2, a classic lightweight network, across all metrics. DenseNet exhibits better performance than ResNet-50 across all seven evaluation metrics. Most significantly, the Swin Transformer architecture emerges as the most effective model, surpassing not only MobileNetV2 but also EfficientNetV2. The empirical results confirm that our Swin Transformer achieves statistically significant improvements in accuracy, precision, recall, specificity, F1-score, Kappa coefficient, and MCC compared to the other models.

Manually checking for defects on industrial parts such as steel surfaces is ineffective, error-prone, and can damage a company’s reputation. Current automated methods often lack accuracy or real-time detection capabilities. Early detection allows for timely corrective action, such as removing defective parts or adjusting production parameters which can streamline the manufacturing process and improve brand reputation, customer satisfaction, and compliance with industry standards. This paper presents MoL-YOLOv7 (MobileNet integrated with attention and loss function to You Only Look Once version 7), a deep learning model for accurate and real-time detection of steel defects. MoL-YOLOv7 modifies the YOLOv7 model by inserting a MobileNet block that reduces computational complexity while maintaining accuracy, allowing for faster detection. Adding the SimAm (Simple Parameter free Attention module) attention mechanism to the MobileNet block refines feature representations for complex tasks such as steel defect detection. Finally, replacing loss functions with EIoU (Effective IoU), WIoU (Wise-IoU), and SIoU (Scylla-IoU) improves the localization accuracy and addresses the class imbalance in the data. The modified model achieves high accuracy and real-time detection, enabling a streamlined defect detection process. Experimental results show that the modified model involving different loss functions used in this work achieves high accuracy, i.e. 0.5% to 3.5% higher than the original model YOLOv7. The superiority and validity of our modified model are demonstrated by comparison with other attention mechanisms and loss functions integrated into YOLOv7, and also on different texture datasets, putting forward a modified method to detect surface defects on steel strips in daily operations.

Brain tumors, a significant health concern, are a leading cause of mortality globally, with an annual projected increase of 5% by the World Health Organization. This work aims to comprehensively analyze the performance of transfer learning methods in identifying the types of brain tumors, with a particular emphasis on the necessity of prompt identification. The study demonstrates how useful it is to use pre-trained models, including models VGG-16, VGG-19, Inception-v3, ResNet-50, DenseNet, and MobileNet—on MRI datasets and used to obtain a precise classification. Using these methods model accuracy and efficiency have been enhanced. The research aims to contribute to improved treatment planning and patient outcomes by implementing optimal methodologies for precise and automated brain tumor analysis, evaluation framework encompasses vital metrics such as confusion matrices, ROC curves, and the achieved Area Under the Curve (AUC) for each approach. The comprehensive methodology outlined in this paper serves as a systematic guide for the implementation and evaluation of brain tumor classification models utilizing deep learning techniques. The integration of visual representations, code snippets, and performance metrics significantly enhances the clarity and understanding of the proposed approach. Among our proposed algorithms, VGG-16 attains the highest accuracy at 97% and consumes only 22% of time as compared to our previous proposed methodology.

This paper presents an efficient deep neural network (DNN) accelerator designed for radio frequency (RF) signal modulation recognition. A novel DNN design optimized for mobile applications is demonstrated by combining MobileNetV3-based DNN with a ternary weight quantization. We also propose a new training method called decaying weight training to overcome the performance degradation due to quantization. The effect of the ternary weight quantization is demonstrated with a co-analysis of the classification accuracy and the physical design. The physical design analysis is based on the Application Specific Integrated Circuit (ASIC), and the results show that the ternary weight quantization with the proposed training method minimizes the impact of the quantization while increasing the allowable clock frequency and reducing hardware cost significantly. We also implement the hardware design dedicated to the ternary weight networks to reduce the required number of the multiply and accumulate (MAC) engines. The hardware design is verified on FPGA and the ternary weight-based DNN shows the feasibility of reducing the hardware cost significantly.

This paper presents a comprehensive study on fault identification in Hall sensors within Brushless Direct Current (BLDC) motor drives using neural networks. Detecting these faults is critical for optimizing motor performance, enhancing energy efficiency, and ensuring overall reliability. Our objective is to propose an accurate, compact, and efficient fault detection neural network model for real-time monitoring and swift responses to Hall sensor faults. Conventional methods are often computationally demanding and fail to detect subtle faults, leading to significant performance decline and potential catastrophic failures. To address this, we leverage MobileNet-based compact models and compare them with state-of-the-art neural network models to identify the most accurate, lightweight, and fast inference option. Our findings demonstrate that MobileNet models excel in detecting Hall sensor faults, achieving over 90% accuracy with significantly fewer parameters (less than five million) and an impressive inference time of under 25 milliseconds. This highlights MobileNet as a robust and efficient choice for Hall sensor fault detection in BLDC motor drives.

Traffic accidents continue to be a significant cause of fatalities, injuries, and considerable disruptions on our highways. Understanding the underlying factors behind these incidents is crucial for improving safety on road networks. While recent studies have highlighted the usefulness of predictive modeling in uncovering factors leading to accidents, there remains a gap in explaining the inner workings of complex machine learning and deep learning models and how various features influence accident prediction. This lack of transparency may lead to these models being perceived as black boxes, potentially undermining trust in their findings among stakeholders. The primary aim of this research is to develop predictive models using diverse transfer learning techniques and shed light on the most influential factors using Shapley values. In predicting injury severity in accidents, we employ Multilayer Perceptron (MLP), Convolutional Neural Network (CNN), Long Short-Term Memory (LSTM), Residual Networks (ResNET), EfficientNetB4, InceptionV3, Extreme Inception (Xception), Visual Geometry Group (VGG19), AlexNet, and MobileNet. Among these models, MobileNet emerges with the highest accuracy at 0.9817. Furthermore, by comprehending how different features impact accident prediction models, researchers can deepen their understanding of the factors contributing to accidents and devise more effective interventions for their prevention.

Today, specific convolution neural network (CNN) models assigned to specific tasks are often used. In this article, the authors explored three models: MobileNet, EfficientNetB0, and InceptionV3 combined. The authors were interested in investigating how quickly an artificial intelligence model can be taught with limited computer resources. Three types of training bases were investigated, starting with a simple base verifying five colours, then recognizing two different orthogonal elements, followed by more complex images from different families. This research aimed to demonstrate the capabilities of the models based on training base parameters such as the number of images and epoch types. Architectures proposed by the authors in these cases were chosen based on simulation studies conducted on a virtual machine with limited hardware parameters. The proposals present the advantages and disadvantages of the different models based on the TensorFlow and Keras libraries in the Jupiter environment based on the Python programming language. An artificial intelligence model with a combination of MobileNet, proposed by Siemens, and Efficient and Inception, selected by the authors, allows for further work to be conducted on image classification, but with limited computer resources for industrial implementation on a programmable logical controller (PLC). The study showed a 90% success rate, with a learning time of 180 s. [ABSTRACT FROM AUTHOR]

Breast cancer ranks among the most prevalent forms of malignancy and foremost cause of death by cancer worldwide. It is not preventable. Early and precise detection is the only remedy for lowering the rate of mortality and improving the probability of survival for victims. In contrast to present procedures, thermography aids in the early diagnosis of cancer and thereby saves lives. But the accuracy experiences detrimental impact by low sensitivity for small and deep tumours and the subjectivity by physicians in interpreting the images. Employing deep learning approaches for cancer detection can enhance the efficacy. This study explored the utilization of thermography in early identification of breast cancer with the use of a publicly released dataset known as the DMR-IR dataset. For this purpose, we employed a novel approach that entails the utilization of a pre-trained MobileNetV2 model and fine tuning it through transfer learning techniques. We created three models using MobileNetV2: one was a baseline transfer learning model with weights trained from ImageNet dataset, the second was a fine-tuned model with an adaptive learning rate, and the third utilized early stopping with callbacks during fine-tuning. The results showed that the proposed methods achieved average accuracy rates of 85.15%, 95.19%, and 98.69%, respectively, with various performance indicators such as precision, sensitivity and specificity also being investigated. [ABSTRACT FROM AUTHOR]

The field of visually impaired assistive technology looks for novel approaches to enhance independence and navigation. In this field, systems have to reliably identify and transmit environmental data in order to facilitate visually impaired users' safe and effective navigation. Developing a sophisticated framework for assistive technology that significantly enhances visually impaired navigation is the goal of this research. Make object detection and environmental awareness more efficient, dependable, and intuitive. This study introduces VISISENSE, "A Comprehensive IoT-Based Assistive Technology System for Enhanced Navigation Support for the Visually Impaired." VISISENSE is an IoT-based system with multiple components that enhances object detection. For primary environmental sensing, a handstick with implant sensors, a visual capture and transmission unit for processing visual data, and edge computing for object detection and classification are used. The system makes use of the R-CNN global computer vision model hosted on a cloud server, Mobinet computer vision models, and Logistic Regression with Iterative Learning. VISISENSE's effectiveness is demonstrated by a performance analysis of its object detection accuracy, processing speed, resource utilization, energy consumption, latency, and false positive rate. In all of these categories, VISISENSE performs better than Smart Stick and Smart Navigation. The data includes the fastest processing time of 17 ms, the most efficient resource utilization of 41%, and object detection accuracy of up to 99% at 2 Mbps load. Across all load conditions, VISISENSE has the lowest false positive rate, energy consumption, and latency. The VISISENSE assistive technology system is developed for the visually impaired. Its excellent object detection and navigation accuracy, speed, and efficiency enhance user experience and hold the potential to increase the independence and quality of life for visually impaired people. This research contributes to a significant advancement in assistive devices--smart, responsive technologies for the visually impaired. [ABSTRACT FROM AUTHOR]

The ear is a visible organ with a unique structure for each person. As a result, it can be used as a biometric to circumvent the constraints of person identification. Deep learning methods like You Only Look Once (YOLO) and MobileNet have recently significantly aided real-time biometric recognition. As a result, in this paper, we approach identifying a person using YOLOV3, YOLOV5, MobileNet-SSDV1, and MobileNet-SSDV2 deep learning algorithms using their ear biometrics. The used ear biometric is a standard dataset (EarVN1.0 Dataset) from 164 individuals with a total of 27,592 images. We chose 10 people at random, totaling 2057 pictures. Of these, 85% were used for training, 5% for validation, and 10% for testing. The performance of the algorithms is determined based on their accuracy and how smoothly the ear of a person is detected. The training accuracy of the algorithms is thresholded at 99.87%. MobileNet-SSDV1, MobileNet-SSDV2, YOLOV3, and YOLOV5 have testing accuracy that is 88%, 91%, 95%, and 96%, respectively. We concluded that the YOLOV5 model outperforms the others in terms of accuracy and size (16MB) for person identification using ear biometrics. [ABSTRACT FROM AUTHOR]

This study proposes a drone system with visual identification and tracking capabilities to address the issue of limited communication bandwidth for drones. This system can lock onto a target during flight and transmit its simple features to the ground station, thereby reducing communication bandwidth demands. RealFlight is used as the simulation environment to validate the proposed drone algorithm. The core components of the system include DeepSORT and MobileNet lightweight models for target tracking. The designed fuzzy controller enables the system to adjust the drone's motors, gradually moving the locked target to the center of the frame and maintaining continuous tracking. Additionally, this study introduces channel and spatial reliability tracking (CSRT) switching from multi-object to single-object tracking and multithreading technology to enhance the system's execution speed. The experimental results demonstrate that the system can accurately adjust the target to the frame's center within approximately 1.5 s, maintaining precision within ±0.5 degrees. On the Jetson Xavier NX embedded platform, the average frame rate (FPS) for the multi-object tracker was only 1.37, with a standard deviation of 1.05. In contrast, the single-object tracker CSRT exhibited a significant improvement, achieving an average FPS of 9.77 with a standard deviation of 1.86. This study provides an effective solution for visual tracking in drone systems that is efficient and conserves communication bandwidth. The validation of the embedded platform highlighted its practicality and performance. [ABSTRACT FROM AUTHOR]

Face recognition technology has been widely used with the WeChat applet on mobile devices; however, facial images are captured on mobile devices and then transmitted to a server for feature extraction and recognition in most existing systems. There are significant security risks related to personal information leakage with these transmissions. Therefore, we propose a face recognition framework for the WeChat applet in which face features are extracted in WeChat by the proposed Face Feature Extraction Model based on Attention Mechanism (FFEM-AM), and only the extracted features are transmitted to the server for recognition. In order to balance the prediction accuracy and model complexity, the structure of the proposed FFEM-AM is lightweight, and Efficient Channel Attention (ECA) was introduced to improve the prediction accuracy. The proposed FFEM-AM was evaluated using a self-built database and the WeChat applet on mobile devices. The experiments show that the prediction accuracy of the proposed FFEM-AM was 98.1%, the running time was less than 100 ms, and the memory cost was only 6.5 MB. Therefore, this demonstrates that the proposed FFEM-AM has high prediction accuracy and can also be deployed with the WeChat applet. [ABSTRACT FROM AUTHOR]

Since the COVID-19 pandemic occurred all over the world, numerous studies were carried out to overcome this problem, including COVID-19 image analysis. An expert analysis based on the Chest X-ray images of COVID-19 determines the progression of the lung condition. Eye visualization and expertise of a radiologist have limitations in handling big cases. This study aims to implement the Convolutional Neural Network (CNN) and MobileNet models as deep learning models to classify chest X-ray images into multiclassification, three categories: COVID-19, normal, and virus. The processes were pre-processing and processing. The pre-processing stage was preparing data, and the processing stage was the implementation model and investigating the best model performance in both convolution and classification in depth-wise convolution and batch normalization. The metrics were accuracy, precision, f1-score, and recall. The CNN results of accuracy, precision, recall, and f1-score respectively were 0.94; 0.99; 0.95; and 0.96. The MobileNet results of the metrics were 0.97; 0.98; 0.99, and 0.99. The MobileNet outperforms the CNN results due to depth-wise convolution and batch normalization. Both models contribute to the faster epoch of the best hyperparameter to achieve loss and accuracy convergence. The models are worth recommending to deployment front-end.

Potato crop has become integral part of our diet due to its wide use in variety of dishes, making it an important food crop. Its importance also stems from the fact that it is one of the cheapest vegetables available throughout the year. This makes it crucial to keep potato prices affordable for developing countries where the majority of the population falls under the middle-income bracket. Consequently, there is a need to develop a robust, effective, and portable technique to detect diseases in potato plant leaves. In this work, an attention-based disease detection technique is proposed. This technique selectively focuses on specific parts of an image which reveal the disease. This technique leverages transfer learning combined with two attention modules: the channel attention module and spatial attention module. By focusing on specific parts of the images, the proposed technique is able to achieve almost similar accuracy with significantly fewer parameters. The proposed technique has been validated using four pre-trained models: DenseNet169, XceptionNet, MobileNet, and VGG16. All of these models are able to achieve almost the same level of training and validation accuracy, around 90–97%, even after reducing the number of parameters by 40–50%. It shows that the proposed technique effectively reduces model complexity without compromising performance.

In recent years, the ability to accurately predict age and gender from facial images has gained significant traction across various fields such as personalized marketing, human–computer interaction, and security surveillance. However, the high computational cost of the current models limits their practicality for real-time applications on resource-constrained devices. This study addressed this challenge by leveraging knowledge distillation to develop lightweight age and gender prediction models that maintain a high accuracy. We propose a knowledge distillation method using teacher bounds for the efficient learning of small models for age and gender. This method allows the student model to selectively receive the teacher model’s knowledge, preventing it from unconditionally learning from the teacher in challenging age/gender prediction tasks involving factors like illusions and makeup. Our experiments used MobileNetV3 and EfficientFormer as the student models and Vision Outlooker (VOLO)-D1 as the teacher model, resulting in substantial efficiency improvements. MobileNetV3-Small, one of the student models we experimented with, achieved a 94.27% reduction in parameters and a 99.17% reduction in Giga Floating Point Operations per Second (GFLOPs). Furthermore, the distilled MobileNetV3-Small model improved gender prediction accuracy from 88.11% to 90.78%. Our findings confirm that knowledge distillation can effectively enhance model performance across diverse demographic groups while ensuring efficiency for deployment on embedded devices. This research advances the development of practical, high-performance AI applications in resource-limited environments.

With the rapid development of the Industrial Internet of Things in rotating machinery, the amount of data sampled by mechanical vibration wireless sensor networks (MvWSNs) has increased significantly, straining bandwidth capacity. Concurrently, the safety requirements for rotating machinery have escalated, necessitating enhanced real-time data processing capabilities. Conventional methods, reliant on experiential approaches, have proven inefficient in meeting these evolving challenges. To this end, a fault detection method for rotating machinery based on mobileNet in MvWSNs is proposed to address these intractable issues. The small and light deep learning model is helpful to realize nearly real-time sensing and fault detection, lightening the communication pressure of MvWSNs. The well-trained deep learning is implanted on the MvWSNs sensor node, an edge computing platform developed via embedded STM32 microcontrollers (STMicroelectronics International NV, Geneva, Switzerland). Data acquisition, data processing, and data classification are all executed on the computing- and energy-constrained sensor node. The experimental results demonstrate that the proposed fault detection method can achieve about 0.99 for the DDS dataset and an accuracy of 0.98 in the MvWSNs sensor node. Furthermore, the final transmission data size is only 0.1% compared to the original data size. It is also a time-saving method that can be accomplished within 135 ms while the raw data will take about 1000 ms to transmit to the monitoring center when there are four sensor nodes in the network. Thus, the proposed edge computing method shows good application prospects in fault detection and control of rotating machinery with high time sensitivity.

Rail surface defects typically serve as early indicators of railway malfunctions, which may compromise the quality and corrosion resistance of rails, thereby endangering the safe operation of trains. The timely detection of defects is essential to ensure the safe operation of railways. To improve the classification accuracy of rail surface defect detection, this paper proposes a rail surface defects detection algorithm based on MobileNet-YOLOv7. By integrating lightweight deep learning algorithms into the engineering application of rail surface defect detection, a MobileNetV3 lightweight network is used as the backbone network for YOLOv7 to enhance both speed and accuracy in complex defect extraction. Subsequently, the efficient intersection over union (EIOU) loss function is utilized as the positional loss function to bolster system resilience. Finally, the k-means++ clustering algorithm is applied to obtain new anchor boxes. The experimental results demonstrate the effectiveness of the proposed method, achieving superior detection accuracy compared with traditional algorithms.

The COVID-19 disease outbreak resulted in a worldwide pandemic. Currently, the reverse transcription-polymerase chain reaction (RT-PCR), which relies on nasopharyngeal swabs to examine the existence of the ribonucleic acid (RNA) of SARS-CoV-27, is still a popular approach to testing for the disease. Despite the high level of specificity of testing with RT-PCR, the sensitivity of the method could be relatively low, and there is significant variability in efficacy depending on different sampling methods and the time of occurrence of symptoms. It is therefore essential for us to develop a machine-learning algorithm that can analyze computerized tomography images to detect the presence of COVID-19. Besides COVID-19, lung computerized tomography (CT) scan images can detect many other diseases, such as lung cancer, pneumonia, etc. This paper deals with the implementation of an algorithm that takes lung CT scans and lung X-ray images as input and predicts a list of probable diseases and possible diagnoses that infect the lungs. Machine learning algorithms will be able to predict disease by scanning the tiniest of regions easily missed by the human eye. This paper presents a survey of various machine learning algorithms that aid in detecting multiple diseases in lung CT scan images. Apart from the study of standard algorithms best suited for COVID-19 detection, this paper also includes recent trends. One of the major recent trends that can be incorporated into COVID-19 detection is TinyML. Tiny ML is an emerging area in machine learning algorithms that can be used to detect multiple diseases in lung CT scan images with better accuracy and in less time. This tool can aid doctors in their diagnosis and treatment of patients and help increase the efficiency of the treatment process. While understanding the features and mapping them using a hidden layer, there is a probability of compressing the dataset, as well as the model to process and classify the low-bit images in real-time using TinyML. [ABSTRACT FROM AUTHOR]

Hand gesture recognition (HGR) is a challenging and fascinating research topic in computer vision with numerous daily life applications. In HGR, computers aim to identify and classify hand gestures. The limited diversity of the dataset used in HGR is due to the limited number of hand gesture demonstrators, acquisition environments, and hand pose variations despite previous efforts. Geometric image augmentations are commonly used to address these limitations. These augmentations include scaling, translation, rotation, flipping, and image shearing. However, research has yet to focus on identifying the best geometric transformations for augmenting the HGR dataset. This study employed three commonly utilized pre-trained models for image classification tasks, namely ResNet50, MobileNetV2, and InceptionV3. The system's performance was evaluated on five static HGR datasets: DLSI, HG14, ArabicASL, MU HandImages ASL, and Sebastian Marcell. The experimental results demonstrate that many geometric transformations are unnecessary for HGR image augmentation. Image shearing and horizontal flipping are the most influential transformations for augmenting the HGR dataset and achieving better classification performance. Moreover, ResNet50 outperforms MobileNetV2 and InceptionV3 for static HGR. [ABSTRACT FROM AUTHOR]

The 2019 new coronavirus illness (COVID-19) is an international public health emergency. Our social and healthcare systems are under a great deal of strain as a result of the daily increase in infection rates and fatalities. Doctors typically perform a chest Xray to identify the diseased areas of the lungs since pneumonia is a common type of infection that spreads in the lungs. In this paper, we propose a Convolution Neural Network via the li regularization model to detect COVID-19 patients using chest X-Ray images. Due to the lack of the COVID-19 benchmark dataset, we use deep learning techniques to identify the best pre-trained CNN model for this job by comparing 15 models. The suggested algorithm was tested on 1316 photos (116 COVID-19 cases, 328 healthy controls, and 872 pneumonia cases), with 66% for training, 17% for validation, and 17% for testing. The classification accuracy, loss, valueaccuracy, and value-loss values obtained by the suggested technique are 0.9912, 0.0187, 0.1119, and 0.9506 respectively. Additionally, the model effectively decreases training loss while boosting accuracy. The results show that proposed procedures are more effective than existing ones at identifying COVID-19 cases from chest X-ray pictures. [ABSTRACT FROM AUTHOR]

Emotion recognition is crucial in artificial intelligence, particularly in the domain of human–computer interaction. The ability to accurately discern and interpret emotions plays a critical role in helping machines to effectively decipher users' underlying intentions, allowing for a more streamlined interaction process that invariably translates into an elevated user experience. The recent increase in social media usage, as well as the availability of an immense amount of unstructured data, has resulted in a significant demand for the deployment of automated emotion recognition systems. Artificial intelligence (AI) techniques have emerged as a powerful solution to this pressing concern in this context. In particular, the incorporation of multimodal AI-driven approaches for emotion recognition has proven beneficial in capturing the intricate interplay of diverse human expression cues that manifest across multiple modalities. The current study aims to develop an effective multimodal emotion recognition system known as MM-EMOR in order to improve the efficacy of emotion recognition efforts focused on audio and text modalities. The use of Mel spectrogram features, Chromagram features, and the Mobilenet Convolutional Neural Network (CNN) for processing audio data are central to the operation of this system, while an attention-based Roberta model caters to the text modality. The methodology of this study is based on an exhaustive evaluation of this approach across three different datasets. Notably, the empirical findings show that MM-EMOR outperforms competing models across the same datasets. This performance boost is noticeable, with accuracy gains of an impressive 7% on one dataset and a substantial 8% on another. Most significantly, the observed increase in accuracy for the final dataset was an astounding 18%. [ABSTRACT FROM AUTHOR]

Using neural networks on low-power mobile systems can aid in controlling pests while preserving beneficial species for crops. However, low-power devices require simplified neural networks, which may lead to reduced performance. This study was focused on developing an optimized deep-learning model for mobile devices for detecting corn pests. We propose a two-step transfer learning approach to enhance the accuracy of two versions of the MobileNet SSD network. Five beetle species (Coleoptera), including four harmful to corn crops (belonging to genera Anoxia, Diabrotica, Opatrum and Zabrus), and one beneficial (Coccinella sp.), were selected for preliminary testing. We employed two datasets. One for the first transfer learning procedure comprises 2605 images with general dataset classes 'Beetle' and 'Ladybug'. It was used to recalibrate the networks' trainable parameters for these two broader classes. Furthermore, the models were retrained on a second dataset of 2648 images of the five selected species. Performance was compared with a baseline model in terms of average accuracy per class and mean average precision (mAP). MobileNet-SSD-v2-Lite achieved an mAP of 0.8923, ranking second but close to the highest mAP (0.908) obtained by MobileNet-SSD-v1 and outperforming the baseline mAP by 6.06%. It demonstrated the highest accuracy for Opatrum (0.9514) and Diabrotica (0.8066). Anoxia it reached a third-place accuracy (0.9851), close to the top value of 0.9912. Zabrus achieved the second position (0.9053), while Coccinella was reliably distinguished from all other species, with an accuracy of 0.8939 and zero false positives; moreover, no pest species were mistakenly identified as Coccinella. Analyzing the errors in the MobileNet-SSD-v2-Lite model revealed good overall accuracy despite the reduced size of the training set, with one misclassification, 33 non-identifications, 7 double identifications and 1 false positive across the 266 images from the test set, yielding an overall relative error rate of 0.1579. The preliminary findings validated the two-step transfer learning procedure and placed the MobileNet-SSD-v2-Lite in the first place, showing high potential for using neural networks on real-time pest control while protecting beneficial species. [ABSTRACT FROM AUTHOR]

In this paper, research was conducted on anomaly detection of wheel flats. In the railway sector, conducting tests with actual railway vehicles is challenging due to safety concerns for passengers and maintenance issues as it is a public industry. Therefore, dynamics software was utilized. Next, STFT (short-time Fourier transform) was performed to create spectrogram images. In the case of railway vehicles, control, monitoring, and communication are performed through TCMS, but complex analysis and data processing are difficult because there are no devices such as GPUs. Furthermore, there are memory limitations. Therefore, in this paper, the relatively lightweight models LeNet-5, ResNet-20, and MobileNet-V3 were selected for deep learning experiments. At this time, the LeNet-5 and MobileNet-V3 models were modified from the basic architecture. Since railway vehicles are given preventive maintenance, it is difficult to obtain fault data. Therefore, semi-supervised learning was also performed. At this time, the Deep One Class Classification paper was referenced. The evaluation results indicated that the modified LeNet-5 and MobileNet-V3 models achieved approximately 97% and 96% accuracy, respectively. At this point, the LeNet-5 model showed a training time of 12 min faster than the MobileNet-V3 model. In addition, the semi-supervised learning results showed a significant outcome of approximately 94% accuracy when considering the railway maintenance environment. In conclusion, considering the railway vehicle maintenance environment and device specifications, it was inferred that the relatively simple and lightweight LeNet-5 model can be effectively utilized while using small images. [ABSTRACT FROM AUTHOR]

To improve the practical effect of digital technology (DT) in electronic education platform (ECP) in sports training, this study proposes using OpenPose as the main method of posture matching in sports teaching. MobileNet was introduced to replace the backbone network in OpenPose to improve the efficiency of the model. At the same time, self-attention mechanism (SAM) is introduced to improve the accuracy of the model. In the performance verification results, the proposed algorithm has a mean mean absolute error (MAE) of 0.793 and a root mean square error (RMSE) of 0.628. The accuracy of the proposed algorithm reached a maximum of 0.893 in the test set. The accuracy of the proposed algorithm exceeds 90% in the training set. In the practice of physical education (PE) and training, the absolute error between the actual value and the predicted value does not exceed 1.5 cm, while the relative error does not exceed 1.2%. The minimum absolute error and relative error are 0.01 cm and 0.2%, respectively. The proposed algorithm has high prediction effect and performance, and can be applied to PE training. [ABSTRACT FROM AUTHOR]

Researchers have explored various potential indicators of ASD, including changes in brain structure and activity, genetics, and immune system abnormalities, but no definitive indicator has been found yet. Therefore, this study aims to investigate ASD indicators using two types of magnetic resonance images (MRI), structural (sMRI) and functional (fMRI), and to address the issue of limited data availability. Transfer learning is a valuable technique when working with limited data, as it utilizes knowledge gained from a pre-trained model in a domain with abundant data. This study proposed the use of four vision transformers namely ConvNeXT, MobileNet, Swin, and ViT using sMRI modalities. The study also investigated the use of a 3D-CNN model with sMRI and fMRI modalities. Our experiments involved different methods of generating data and extracting slices from raw 3D sMRI and 4D fMRI scans along the axial, coronal, and sagittal brain planes. To evaluate our methods, we utilized a standard neuroimaging dataset called NYU from the ABIDE repository to classify ASD subjects from typical control subjects. The performance of our models was evaluated against several baselines including studies that implemented VGG and ResNet transfer learning models. Our experimental results validate the effectiveness of the proposed multi-slice generation with the 3D-CNN and transfer learning methods as they achieved state-of-the-art results. In particular, results from 50-middle slices from the fMRI and 3D-CNN showed a profound promise in ASD classifiability as it obtained a maximum accuracy of 0.8710 and F1-score of 0.8261 when using the mean of 4D images across the axial, coronal, and sagittal. Additionally, the use of the whole slices in fMRI except the beginnings and the ends of brain views helped to reduce irrelevant information and showed good performance of 0.8387 accuracy and 0.7727 F1-score. Lastly, the transfer learning with the ConvNeXt model achieved results higher than other transformers when using 50-middle slices sMRI along the axial, coronal, and sagittal planes. [ABSTRACT FROM AUTHOR]

Identifying sands is an important requirement in many geotechnical exploration projects. Knowledge of the sand type can help in estimating the physical and mechanical properties of a soil. Previous studies revealed that machine learning approaches using neural networks (NN) can correctly identify up to 75% of individual sand particles using size and shape descriptors. This study explores the efficacy of deep learning methods for automatically classifying sand types from individual images of sand particles. Dynamic image analysis (DIA) was employed to generate a large data set of sand particle images, which were then used for training a deep learning model known as convolutional neural networks (CNN). The analysis was based on 40,000 binary particle images for twenty types of sand. The work demonstrates that computer vision has a remarkable ability to automatically classify 64% of individual sand particles among 20 types of sand, the accuracy for sand clusters can reach up to 100%, when a CNN model augmented with size and shape data was employed. The effects of domain size, size and shape parameters, and the selected CNN model on the classification accuracy were also investigated. The results demonstrate that sand classification using a larger number of sand types resulted in a lower classification accuracy. Classification accuracy of individual particles achieved using CNN were 10%–15% better than those achieved using NN. CNN can automatically and adaptively learn the spatial hierarchy of features which is superior to the handcrafted size and shape parameters, used by NN for identifying sand particles. The study suggests that classification accuracy benefits from data augmentation use of more particle orientations, even at the cost of trimming some particles, and suffers from reduction in image resolution. While model training requires a lot of computational work, a pretrained CNN model may potentially be tuned to run on mobile phones, which points to the potential for real-time field deployment to enable automatic soil classification on-site. [ABSTRACT FROM AUTHOR]

Oral lesions are a prevalent manifestation of oral disease, and the timely identification of oral lesions is imperative for effective intervention. Fortunately, deep learning algorithms have shown great potential for automated lesion detection. The primary aim of this study was to employ deep learning-based image classification algorithms to identify oral lesions. We used three deep learning models, namely VGG19, DeIT, and MobileNet, to assess the efficacy of various categorization methods. To evaluate the accuracy and reliability of the models, we employed a dataset consisting of oral pictures encompassing two distinct categories: benign and malignant lesions. The experimental findings indicate that VGG19 and MobileNet attained an almost perfect accuracy rate of 100 % , while DeIT achieved a slightly lower accuracy rate of 98.73%. The results of this study indicate that deep learning algorithms for picture classification demonstrate a high level of effectiveness in detecting oral lesions by achieving 100 % for VGG19 and MobileNet and 98.73 % for DeIT. Specifically, the VGG19 and MobileNet models exhibit notable suitability for this particular task. [ABSTRACT FROM AUTHOR]

Sign languages are complex, but there are ongoing research efforts in engineering and data science to recognize, understand, and utilize them in real-time applications. Arabic sign language recognition (ArSL) has been examined and applied using various traditional and intelligent methods. However, there have been limited attempts to enhance this process by utilizing pretrained models and large-sized vision transformers designed for image classification tasks. This study aimed to create robust transfer learning models trained on a dataset of 54,049 images depicting 32 alphabets from an ArSL dataset. The goal was to accurately classify these images into their corresponding Arabic alphabets. This study included two methodological parts. The first one was the transfer learning approach, wherein we utilized various pretrained models namely MobileNet, Xception, Inception, InceptionResNet, DenseNet, and BiT, and two vision transformers namely ViT, and Swin. We evaluated different variants from base-sized to large-sized pretrained models and vision transformers with weights initialized from the ImageNet dataset or otherwise randomly. The second part was the deep learning approach using convolutional neural networks (CNNs), wherein several CNN architectures were trained from scratch to be compared with the transfer learning approach. The proposed methods were evaluated using the accuracy, AUC, precision, recall, F1 and loss metrics. The transfer learning approach consistently performed well on the ArSL dataset and outperformed other CNN models. ResNet and InceptionResNet obtained a comparably high performance of 98%. By combining the concepts of transformer-based architecture and pretraining, ViT and Swin leveraged the strengths of both architectures and reduced the number of parameters required for training, making them more efficient and stable than other models and existing studies for ArSL classification. This demonstrates the effectiveness and robustness of using transfer learning with vision transformers for sign language recognition for other low-resourced languages. [ABSTRACT FROM AUTHOR]

Background: COVID-19 is a highly contagious respiratory disease with multiple mutant variants, an asymptotic nature in patients, and with potential to stay undetected in common tests, which makes it deadlier, more transmissible, and harder to detect. Regardless of variants, the COVID-19 infection shows several observable anomalies in the computed tomography (CT) scans of the lungs, even in the early stages of infection. A quick and reliable way of detecting COVID-19 is essential to manage the growing transmission of COVID-19 and save lives. Objective: This study focuses on developing a deep learning model that can be used as an auxiliary decision system to detect COVID-19 from chest CT-scan images quickly and effectively. Methods: In this research, we propose a MobileNet-based transfer learning model to detect COVID-19 in CT-scan images. To test the performance of our proposed model, we collect three publicly available COVID-19 CT-scan datasets and prepare another dataset by combining the collected datasets. We also implement a mobile application using the model trained on the combined dataset, which can be used as an auxiliary decision system for COVID-19 screening in real life. Results: Our proposed model achieves a promising accuracy of 96.14% on the combined dataset and accuracy of 98.75%, 98.54%, and 97.84% respectively in detecting COVID-19 samples on the collected datasets. It also outperforms other transfer learning models while having lower memory consumption, ensuring the best performance in both normal and low-powered, resource-constrained devices. Conclusion: We believe, the promising performance of our proposed method will facilitate its use as an auxiliary decision system to detect COVID-19 patients quickly and reliably. This will allow authorities to take immediate measures to limit COVID-19 transmission to prevent further casualties as well as accelerate the screening for COVID-19 while reducing the workload of medical personnel. [ABSTRACT FROM AUTHOR]