Embedded system vehicle based on multi-sensor fusion

As intelligent driving vehicles came out of concept into people’s life, the combination of safe driving and artificial intelligence becomes the new direction of future transportation development. Autonomous driving technology is developing based on control algorithms and model recognitions. In this paper, a cloud-based interconnected multi-sensor fusion autonomous vehicle system is proposed that uses deep learning (YOLOv4) and improved ORB algorithms to identify pedestrians, vehicles, and various traffic signs. A cloud-based interactive system is built to enable vehicle owners to master the situation of their vehicles at any time. In order to meet multiple application of automatic driving vehicles, the environment perception technology of multi-sensor fusion processing has broadened the uses of automatic driving vehicles by being equipped with automatic speech recognition (ASR), vehicle following mode and road patrol mode. These functions enable automatic driving to be used in applications such as agricultural irrigation, road firefighting and contactless delivery under new coronavirus outbreaks. Finally, using the embedded system equipment, an intelligent car was built for experimental verification, and the overall recognition accuracy of the system was over 96%. Author

COVID-19 risk reduce based YOLOv4-P6-FaceMask detector and DeepSORT tracker.

Wearing masks in public areas is one of the effective protection methods for people. Although it is essential to wear the facemask correctly, there are few research studies about facemask detection and tracking based on image processing. In this work, we propose a new high performance two stage facemask detector and tracker with a monocular camera and a deep learning based framework for automating the task of facemask detection and tracking using video sequences. Furthermore, we propose a novel facemask detection dataset consisting of 18,000 images with more than 30,000 tight bounding boxes and annotations for three different class labels namely respectively: face masked/incorrectly masked/no masked. We based on Scaled-You Only Look Once (Scaled-YOLOv4) object detection model to train the YOLOv4-P6-FaceMask detector and Simple Online and Real-time Tracking with a deep association metric (DeepSORT) approach to tracking faces. We suggest using DeepSORT to track faces by ID assignment to save faces only once and create a database of no masked faces. YOLOv4-P6-FaceMask is a model with high accuracy that achieves 93% mean average precision, 92% mean average recall and the real-time speed of 35 fps on single GPU Tesla-T4 graphic card on our proposed dataset. To demonstrate the performance of the proposed model, we compare the detection and tracking results with other popular state-of-the-art models of facemask detection and tracking.

Physical Distancing Monitoring Application with Zone-Based Calibration through YOLOv4 Model

Physical Distancing is one of the minimum health protocols where two persons should be at least 1.5 meters apart to lessen the risk of transmission of COVID-19. The study aims to design a real-time monitoring system that detects violations on physical Distancing by applying the You Only Look Once version 4 computer vision model. The program detects the pairwise distance between two persons in a frame and indicates whether they comply with the minimum 1.5 distance between persons. The video frame comprises zone 1 being the farthest from the camera, zone 2, and zone 3 being the nearest from the camera. The program calculates the Euclidean distance between persons and generates a pixel value converted to a metric value by a scale multiplier. The scaling multiplier varies depending on the zone at which the location of the detected person is. The mean absolute error of the distance predicted by the program is at 7.8 centimeters, 5.73 centimeters, and 5.21 centimeters at zones 1, 2, and 3, respectively. The physical distancing detector achieved 95.84% accuracy and 97.08% precision upon evaluating through the confusion matrix.

Covid Analysis using Recurrent Neural Network

The examination of medical images has benefited greatly from the use of artificial intelligence. In contrast to deep learning systems, which do feature extraction automatically and without human interaction, traditional computer vision methods rely on manually produced features that are particular to a certain domain. Having access to medical information for automated analysis is another major factor driving the trend towards deep learning. Chest x-ray pictures are processed in order to segment the lungs and identify diseases in this thesis. Due to its cheap cost, ease of capture, and non-invasive nature, chest x-ray is the most often used medical imaging technology. However, automatic diagnosis in chest x-rays is difficult due to (1) the presence of the rib-cage and clavicle bones, which can obscure abnormalities that are located beneath them, and (2) the fuzzy intensity transitions near the lung and heart, dense abnormalities, rib-cages, and clavicle bones, which make the identification of lung contours subtle. In x-ray image processing, the Convolutional Neural Network (CNN) is the most often used deep learning architecture. Because to the enormous number of parameters in deep CNN architectures, intensive computing resources are required to train these models. Additionally, chest x-ray datasets are often rather tiny, and there is always the risk of overfitting when developing a model. In this dissertation, we propose five convolutional neural networks (CNNs) to identify illness and segment the lungs in chest x-rays. New Line, New Line In the first research paper, an adaptive lightweight convolutional neural network (ALCNN) is created to detect pneumothorax with few parameters. The model readjusts the feature calibration channel-wise using the convolutional layer and attention mechanism. The suggested model outperformed state-of-the-art deep models trained using three different transfer learning methods. One notable aspect of the suggested model is that it requires ten times less parameters than the best deep models currently available. The second paper suggests the FocusCovid methodology for identifying COVID-19. © 2023 IEEE.

Robust Classification and Detection of Big Medical Data Using Advanced Parallel K-Means Clustering, YOLOv4, and Logistic Regression.

Big-medical-data classification and image detection are crucial tasks in the field of healthcare, as they can assist with diagnosis, treatment planning, and disease monitoring. Logistic regression and YOLOv4 are popular algorithms that can be used for these tasks. However, these techniques have limitations and performance issue with big medical data. In this study, we presented a robust approach for big-medical-data classification and image detection using logistic regression and YOLOv4, respectively. To improve the performance of these algorithms, we proposed the use of advanced parallel k-means pre-processing, a clustering technique that identified patterns and structures in the data. Additionally, we leveraged the acceleration capabilities of a neural engine processor to further enhance the speed and efficiency of our approach. We evaluated our approach on several large medical datasets and showed that it could accurately classify large amounts of medical data and detect medical images. Our results demonstrated that the combination of advanced parallel k-means pre-processing, and the neural engine processor resulted in a significant improvement in the performance of logistic regression and YOLOv4, making them more reliable for use in medical applications. This new approach offers a promising solution for medical data classification and image detection and may have significant implications for the field of healthcare.

YOLO KeepSafe: Utilization of YOLOv4-Based Algorithm to Employ a Real-time People Detection and Social Distance Monitoring System

This paper outlines the process used to create the social distance detection system - YOLO KeepSafe. The researchers discuss several calibration procedures, investigate the difficulties that arise when used in real-time situations, and offer potential solutions. © 2022 IEEE.

Covid-19 Control: Face Mask Detection Using Deep Learning for Balanced and Unbalanced Dataset

Facemask wearing is becoming a norm in our daily lives to curb the spread of Covid-19. Ensuring facemasks are worn correctly is a topic of concern worldwide. It could go beyond manual human control and enforcement, leading to the spread of this deadly virus and many cases globally. The main aim of wearing a facemask is to curtail the spread of the covid-19 virus, but the biggest concern of most deep learning research is about who is wearing the mask or not, and not who is incorrectly wearing the facemask while the main objective of mask wearing is to prevent the spread of the covid-19 virus. This paper compares three state-of-the- art object detection approaches: Haarcascade, Multi-task Cascaded Convolutional Networks (MTCNN), and You Only Look Once version 4 (YOLOv4) to classify who is wearing a mask, who is not wearing a mask, and most importantly, who is incorrectly wearing the mask in a real-time video stream using FPS as a benchmark to select the best model. Yolov4 got about 40 Frame Per Seconds (FPS), outperforming Haarcascade with 16 and MTCNN with 1.4. YOLOv4 was later used to compare the two datasets using Intersection over Union (IoU) and mean Average Precision (mAP) as a comparative measure;dataset2 (balanced dataset) performed better than dataset1 (unbalanced dataset). Yolov4 model on dataset2 mapped and detected images of masks worn incorrectly with one correct class label rather than giving them two label classes with uncertainty in dataset1, this work shows the advantage of having a balanced dataset for accuracy. This work would help decrease human interference in enforcing the COVID-19 face mask rules and create awareness for people who do not comply with the facemask policy of wearing it correctly. Hence, significantly reducing the spread of COVID-19.

Research and Optimization of a Lightweight Refined Mask-Wearing Detection Algorithm Based on an Attention Mechanism

To address the current problems of the incomplete classification of mask-wearing detection data, small-target miss detection, and the insufficient feature extraction capabilities of lightweight networks dealing with complex faces, a lightweight method with an attention mechanism for detecting mask wearing is presented in this paper. This study incorporated an "incorrect_mask” category into the dataset to address incomplete classification. Additionally, the YOLOv4-tiny model was enhanced with a prediction feature layer and feature fusion execution, expanding the detection scale range and improving the performance on small targets. A CBAM attention module was then introduced into the feature enhancement network, which re-screened the feature information of the region of interest to retain important feature information and improve the feature extraction capabilities. Finally, a focal loss function and an improved mosaic data enhancement strategy were used to enhance the target classification performance. The experimental results of classifying three objects demonstrate that the lightweight model's detection speed was not compromised while achieving a 2.08% increase in the average classification precision, which was only 0.69% lower than that of the YOLOv4 network. Therefore, this approach effectively improves the detection effect of the lightweight network for mask-wearing.

Outdoor Social Distancing Violation System Detection Using YOLO Algorithm

The government issued orders to implement social distancing or physical distancing. Social distancing is a method of maintaining a distance of at least one meter from other people. This is useful for reducing/preventing disease transmission (virus) and reducing the chain of the spread of covid-19. So the hospital can provide optimal service. For this reason, this research is structured to create a system that can detect violations of social distancing in an open place. This system uses the You Only Look Once (YOLO) algorithm. The developed system uses a pre-Trained Yolov4 model to detect 80 object classes. Testing of this system is carried out based on several scenarios. The system is programmed using Python, with tools for coding Microsoft Visual Studio Code and Anaconda. The best result from creating the detection mode is obtained from a dataset ratio of 90% train data and 10% test data, with the mean average precision results obtained being 54.11%. © 2022 IEEE.

Social Distance Detector Using Computer Vision and Deep Learning to Combat COVID-19

The study's goal is to create a detector that detects and analyses whether pedestrians or individuals in public gatherings are maintaining social distancing. Drone-shot videos, live webcam feeds, and photographs are all kinds of input for the detector. With no human intervention, Dynamic Detection through live stream provides safety and simplifies monitoring of social distance. The webcam input can be integrated with an external webcam or a drone's camera. Furthermore, the YOLOv4 algorithm is used for the data set for the initial phase ofobject detection, identifying various items in each frame. The recognized objects are narrowed down to humans, and the Euclidian distance between one data point and every other data point is determined The Euclidian distance determines if they are maintaining the minimal distance between them or not by depicting them with a colored border box. Euclidian distance assists in detecting if they are keeping the minimal distance between them or not, as shown by a coloredboundary box, red for unsafe and green for safe, with an indication reflecting the number of people in danger. © 2022 IEEE.

YOLO-Improve Detection Method of Real-Time Mask Wearing

The existing YOLO target detection algorithm, based on one-stage concept, aims for multi-objective detection. But this algorithm performs not well for dual classification detection and add more performance overhead while detection. In order to improve the detection efficiency of dual classification mask wearing during the period of COVID-19, this paper proposes a real-time detection method based on YOLO for detecting the condition of bi-objective mask wearing. The optimization is based on the latest YOLO version 4: firstly, the feedforward input layer of the model is improved to optimize the data augment part and the adaptive image scaling is added to improve the efficiency of the detection for both dual classification and small objectives;secondly, the adaptive anchoring frame is added to replace the activation function so as to reduce the computation and thus improve its efficiency;thirdly, the optimization of Neck and the addition of Focus structure improve the capability of feature fusion and reduce the amount of parameters to raise the efficiency. The experimental results showed that, compared with the existing YOLO version 4, using this method in this paper improves F1 by 0.33% and mAp by 0.71% in the data set. Also, the detection efficiency is also improved significantly under the same experimental condition. (English) [ABSTRACT FROM AUTHOR] 现有的 YOLO 目标检测模型是基于one-stage 思想进行多目标的检测，对于双分类检测其有所不足，并且其检测时性能也消耗较大。为了能够在新冠疫情暴发的特殊时期，提高双分类口罩佩戴的检测精度和检测效率，文中提出了一种基于 YOLO 的双目标口罩佩戴实时检测方法。在最新的 YOLO 第四版的基础上进行了优化。改进了模型的前馈输入层，优化了数据增强部分，添加了自适应图片缩放，对双分类和小目标的检测精度和检测效率都有所提升；添加了自适应锚定框，替换了激活函数，降低方法的计算量从而提高方法的检测效率；Neck部分的优化和 Focus 结构的添加，提高了特征融合能力并且减少参数量达到提速的效果。实验结果表明，与现有的 YOLO 第四版相比，文中的方法在文中数据集中F1提高0.33%，mAp提高了0.71%，并且相同实验环境下检测效率也提升明显。 (Chinese) [ABSTRACT FROM AUTHOR] Copyright of Electronic Science & Technology is the property of Electronic Science & Technology Editorial Office 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.)

A Real-Time Social Distancing and Face Mask Detection System Using Deep Learning

It has been more than two years since the transmission of COVID-19 virus has affected the public health globally. Due to its natural characteristic, the virus is very likely to undergo mutation over time and consistently changes to a new variant with higher severity and transmission rate. The pandemic is expected to prolong with the increment in number of daily cases which leads to why preventive measures like practising distance apart rule and wearing facemask are still mandatory in the long run. This paper is prepared to develop a social distancing model using deep learning for COVID-19 pandemic. The tracking accuracy of the proposed model is discussed in the paper and compared with other deep learning methods as well. The efficiency of the detection model is observed and evaluated by performing quantitative metrics. The monitoring model is trained by implementing YOLOv4 algorithm and has achieved an accuracy of 93.79% with F1-score of 0.87 in detecting person and facemask. The model is applicable for real-time and video detection to monitor social distance violation as an effort to flatten the curve and slow down the transmission rate in the community. © 2023, The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd.

Convolutional Neural Network-based Covid-19 Analysis with Internet of Things

The very hazardous respiratory illness known as COVID-2 (SARS-CoV-2), which is the root cause of the even more serious illness known as COVID-19, was caused by the COVID-2 virus. The COVID-19 virus was identified in Wuhan City, China, in the month of December in 2019. It began in China and then spread to other parts of the world before it was officially classified as a pandemic. It has had a significant impact on day-To-day life, the welfare of people in general, and the economy of the whole globe. It is of the utmost importance, particularly in the beginning stages of treatment, to pinpoint the constructive experiences that are useful at the proper time. The identification of this virus involves a substantial number of tests, each of which takes a certain amount of time;nevertheless, there are currently no other automated tool kits that can be used in their place. X-ray photos of the chest that are obtained via the use of radiology imaging methods may provide significant insight into the COVID-19 infection if they are analysed carefully. An accurate diagnosis of the infection may be obtained via the application of deep learning techniques, which are applied to radiological images and make use of cutting-edge technology such as artificial intelligence. Patients who reside in distant places, where it may not be feasible for them to have rapid access to medical facilities, may benefit from this kind of analysis throughout the course of their therapy. One of the deep learning strategies that are used in the creation of the model that has been proposed is the use of convolutional neural networks. The images of chest X-rays are analysed by these networks to detect whether a person has a positive or negative result for the Covid gene. © 2022 IEEE.

Development of Personnel Epidemic Prevention Monitoring Access Control System

Because of coronavirus variants, it is necessary to pay attention to epidemic prevention measures in the cultivation or product packaging processes. In addition to giving customers more peace of mind when using the products, it also ensures that operators wear masks, work clothes and gloves in the work area. This paper constructs an access control system for personnel epidemic prevention monitoring, which uses IoTtalk [1] to connect IoT devices (such as magnetic reed switches, intelligent switches, RFID readers, and RFID wristbands), utilizes RFID for personnel identification, and employs real-time streaming protocol [2] to take the image of IP Cam for YOLOv4 [3] identification program. The identification program detects whether the personnel is indeed wearing the required equipment. If the personnel is not wearing the required device, the detector will trigger a push broadcast system constructed by LINE Notify to inform the operator for processing. Moreover, we developed an emergency entry mechanism;if an emergency happens, the personnel can trigger the emergency door opening by swiping the card multiple times within a specified time. This function allows the person to enter without wearing the required equipment. © 2022, The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd.

Image Recognition Based on Deep Learning with Thermal Camera Sensing

As the COVID-19 epidemic spread across the globe, people around the world were advised or mandated to wear masks in public places to prevent its spreading further. In some cases, not wearing a mask could result in a fine. To monitor mask wearing, and to prevent the spread of future epidemics, this study proposes an image recognition system consisting of a camera, an infrared thermal array sensor, and a convolutional neural network trained in mask recognition. The infrared sensor monitors body temperature and displays the results in real-time on a liquid crystal display screen. The proposed system reduces the inefficiency of traditional object detection by providing training data according to the specific needs of the user and by applying You Only Look Once Version 4 (YOLOv4) object detection technology, which experiments show has more efficient training parameters and a higher level of accuracy in object recognition. All datasets are uploaded to the cloud for storage using Google Colaboratory, saving human resources and achieving a high level of efficiency at a low cost. © 2023 CRL Publishing. All rights reserved.

Proposing Algorithm for Warning Minimum Distance Violation and Tracking Covid-19 Object Using YOLOV4 Model

Currently, the prevention of the spread of the Covid-19 epidemic is still a matter of concern with many new variants that are more infectious and making it more difficult to prevent it. In addition, several respiratory viral diseases such as influenza A, monkeypox, etc. help promote the management and prevention of epidemics. The paper presents the system using the YOLOV4 object recognition model to identify human objects from videos extracted. To increase accuracy with the desired context, we build a dataset of people and perform training on them. We use the Euclidean algorithm to calculate the distance between bounding box pairs. We then use a physical distance that approximates the pixel and set a threshold. It is possible to determine who has violated the minimum social distance threshold. In addition, we apply a tracking algorithm to be able to detect and trace those who have been in close contact with the cases to be monitored. The system has been performed on video and the accuracy of the model is up to 95.6%. © 2022 IEEE.

Effective face mask and social distance detection with alert system for covid-19 using YOLOv5 model

Globally, numerous preventive measures were taken to treat the COVID-19 epidemic. Face masks and social distancing were two of the most crucial practices for limiting the spread of novel viruses. With YOLOv5 and a pre-Trained framework, we present a novel method of complex mask detection. The primary objective is to detect complex different face masks at higher rates and obtain accuracy of about 94% to 99% on real-Time video feeds. The proposed methodology also aims to implement a structure to detect social distance based on a YOLOv5 architecture for controlling, monitoring, accomplishing, and reducing the interaction of physical communication among people in the day-To-day environment. In order for the framework to be trained for the different crowd datasets from the top, it was trained for the human contrasts. Based on the pixel information and the violation threshold, the Euclidean distance between peoples is determined as soon as the people in the video are spotted. In the results, this social distance architecture is described as providing effective monitoring and alerting. © 2022 The authors and IOS Press.

Face Masked and Unmasked Humans Detection and Tracking in Video Surveillance

Due to the spread of COVID-19, people wearing face masks became a regular occurrence worldwide. Moreover, there are nations where covering one's face is done for religious or cultural reasons, or even wear face masks for convenience. However, current face detection and tracking systems are hindered by face masks as the full facial features are no longer visible and therefore became less effective. In this paper, it is proposed to improve current face detection and long-term tracking technology by extracting the facial features of the top regions of the face, taking into account the eye, eyebrow, and forehead. The methodology contains two models, the face detector and the long-term object tracker. The face detection model uses a joint dataset from ISL-UFMD and MaskedFace-Net. The dataset is used to train a Keras sequential model. The object detection model uses pre-trained YOLOv4 weights and DeepSORT to identify people and uses the tracking-by-detection method to perform long-term tracking throughout the surveillance video. The final face detection model results show a testing accuracy of 93.33% and a loss of 26.92%, which are up to par and comparable with other state-of-the-art models. © 2022 IEEE.

Multi category mask wearing detection based on dynamic weighted category balance loss

Mask wearing in public has become an important measure to control the spread of Coronavirus Disease 2019 (COVID-19). With the prolonged development of the COVID-19 epidemic, the public’s awareness of self-protection has been gradually declining, leading to the increasing tendency of wearing masks incorrectly in public. The existing mask wearing detection methods usually only detect whether the mask is worn, without the detection of non-standard mask wearing scenarios, which is likely to cause cross infection. The current mask datasets lack the image data of non-standard mask wearing. To solve the above problems, on the basis of the existing mask datasets, more non-standard mask wearing images were collected through the Internet and offline, and the Mosaic data enhancement algorithm was improved to expand the data according to the features of face images in the cases of wearing masks. The improved Mosaic data enhancement algorithm could improve the mean average precision (mAP) of the benchmark network YOLOv4 by 2.08%. To address the problem of category imbalance in the dataset after data enhancement, the dynamic weighted balance loss function was proposed. Based on the weight binary cross entropy loss function, the reciprocal of the number of effective samples served as the auxiliary category weight, and dynamic adjustment was performed in each batch under training, thus solving the problems of weak stability, precision oscillation, and unsatisfactory effect when the re-weighting method was directly put to use. The experiment showed that mAP of the improved model reached 91.25%, and the average precision (AP) of non-standard mask wearing reached 91.69%. Compared with such single-stage methods as RetinaNet, Centernet, and Effcientdet, and such two-stage methods as YOLOv3-MobileNetV2 and YOLOv4-MobileNetV2, the improved algorithm exhibits higher detection accuracy and speed. © 2022, Editorial of Board of Journal of Graphics. All rights reserved.

A real-time model for COVID19 face-mask identification with 'YOLOv4'

At the beginning of 2020 WHO declared COVID19 as an epidemic;healthcare industries experts and academicians from worldwide are working in the directions to surveillance the daily behaviors of the citizens to combat the COVID-19 cases. In India, we thank the government for performing its outperformed active measures and spontaneous compliance to follow the policy of wearing masks when moving out to any public places;it entails active real-time monitoring to supervise the citizens by governments. In this process, real-time face-mask identification is a very challenging task of computer vision. And the absence of accurate datasets for this problem is a critical hard problem to solve. To address this bottleneck, we are proposing our real-time deep learning face-mask identification technique with annotated class labels with bounding boxes which have its real-time application to assist the governments to control and prevent the spread of these epidemics in its supervision. Our model is very robust and effective to classify the real-time images and videos for face mask detection with accuracy and average precision. The proposed model substitutes the manual surveillance with the object detection method using YOLOv4 supported on a deep learning approach to monitor the crowd accurately even if they change their respective locations. The experiment identify or classify the object within any dataset to distinguish the images or videos with two class labels such as 'with-mask' and 'without-mask' with approximately 98.26% accuracy, mAP of 68.28%, recall of 77%, and precision of 57%. © 2022 IEEE.