Design and Implementation of Intelligent Truck Based on Azure Kinect

In recent years, due to the impact of COVID-19, the market prospect of non-contact handling has improved and the development potential is huge. This paper designs an intelligent truck based on Azure Kinect, which can save manpower and improve efficiency, and greatly reduce the infection risk of medical staff and community workers. The target object is visually recognized by Azure Kinect to obtain the center of mass of the target, and the GPS and Kalman filter are used to achieve accurate positioning. The 4-DOF robot arm is selected to grasp and transport the target object, so as to complete the non-contact handling work. In this paper, different shapes of objects are tested. The experiment shows that the system can accurately complete the positioning function, and the accuracy rate is 95.56%. The target object recognition is combined with the depth information to determine the distance, and the spatial coordinates of the object centroid are obtained in real time. The accuracy rate can reach 94.48%, and the target objects of different shapes can be recognized. When the target object is grasped by the robot arm, it can be grasped accurately according to the depth information, and the grasping rate reaches 92.67%. © 2023, ICST Institute for Computer Sciences, Social Informatics and Telecommunications Engineering.

Control System for Indoor Safety Measures Using a Faster R-CNN Architecture

This paper presents a control system for indoor safety measures using a Faster R-CNN (Region-based Convolutional Neural Network) architecture. The proposed system aims to ensure the safety of occupants in indoor environments by detecting and recognizing potential safety hazards in real time, such as capacity control, social distancing, or mask use. Using deep learning techniques, the system detects these situations to be controlled, notifying the person in charge of the company if any of these are violated. The proposed system was tested in a real teaching environment at Rey Juan Carlos University, using Raspberry Pi 4 as a hardware platform together with an Intel Neural Stick board and a pair of PiCamera RGB (Red Green Blue) cameras to capture images of the environment and a Faster R-CNN architecture to detect and classify objects within the images. To evaluate the performance of the system, a dataset of indoor images was collected and annotated for object detection and classification. The system was trained using this dataset, and its performance was evaluated based on precision, recall, and F1 score. The results show that the proposed system achieved a high level of accuracy in detecting and classifying potential safety hazards in indoor environments. The proposed system includes an efficiently implemented software infrastructure to be launched on a low-cost hardware platform, which is affordable for any company, regardless of size or revenue, and it has the potential to be integrated into existing safety systems in indoor environments such as hospitals, warehouses, and factories, to provide real-time monitoring and alerts for safety hazards. Future work will focus on enhancing the system's robustness and scalability to larger indoor environments with more complex safety hazards.

Social Distancing Monitoring and Contact Tracing on CCTV Footage

Social distancing is one of the most effective measures to prevent the spread of the COVID-19 disease. Most methods of enforcing this in the Philippines resort to manual methods. As such, a video-based social distancing monitoring tool can help ensure constant enforcement of social distancing due to the availability and up-time of CCTV cameras in various areas. This can be achieved by using object detection and tracking techniques. Object detection can be used to detect people within an area, and tracking can be used to watch people who get into close contact with one another. Contact tracing can also be performed by processing the social distancing measurements and tracking information. This information can be stored to keep a record of who has a high risk of infection based on who they came into contact with and for how long. We introduce a social distancing monitoring and contact tracing framework using the EfficientDet object detector and DeepSORT tracker. This framework is used to monitor social distancing violations and keep a record of violations associated to the tracked people. © 2022 IEEE.

Real-Time and Accurate Monitoring of Social Distance and Contact Tracing during COVID-19 Situation

In recent years, a lot of research works have been done on object detection using various machine learning models. However, not many works have been done on detecting and tracking humans in particular. This study works with the YOLOv4 object detector to detect humans to use the detections for maintaining social distance. For this study, the YOLOv4 model is trained on only one class named 'Person'. This is done to improve the speed of detecting humans in real time scenario with satisfying accuracy of 97% to 99%. These detections are then tracked to build a system for maintaining social distance and alerting the authority if a breach in the social distance is detected. This system can be applied at ticket counters, hospitals, offices, factories etc. It can also be used for maintaining social distance among the students and the teachers in the classroom for their safety. © 2022 IEEE.

Joystick Controlled Autonomous Robotic Arm Using Inverse Kinematics with Object Detection

In 2019, the Novel Coronavirus Disease (COVID-19) was categorized as a pandemic. This disease can be transmitted via droplets on items or surfaces within several hours. Therefore, the researchers aimed to develop a wirelessly controlled robot arm and platform capable of picking up objects detected via object detection. Robot arm movements are done via the use of inverse kinematics. Meanwhile, a custom object detection model that can detect objects of interest will be trained and implemented in this project. To achieve this, the researchers utilize various open-source libraries, microcontrollers, and readily available materials to construct and program the entire system. At the end of this research, the prototype could reliably detect objects of interest, along with a grab-and-dispose success rate of 88%. Instruction data can be properly sent and received, and dual web cam image transfer reaches up to 1.72 frames per second. © 2023 IEEE.

Design of People Count System Based on Object Detection Algorithm

At present, the Covid-19 epidemic is still spreading globally. Although the domestic epidemic has been well controlled, the prevention and control of the epidemic must not be taken lightly. Being able to count the number of people in public places in real time has played a vital role in the prevention and control of the epidemic. Deep learning networks usually cannot be directly deployed on embedded devices with low computing power due to the huge amount of parameters of convolutional neural networks. This article is based on the YOLOv5 object detection algorithm and Jetson Nano embedded platform with TensorRT and C++ accelerating, it can realize the function of counting the number of people in the classroom, on the elevator entrance, and other scenes. © 2022 SPIE.

Technology-Driven Implementation of Smart Entrances in Public Places during the COVID-19

This research proposes a smart entrance system to cope with the COVID-19 pandemic in public places. The system can help automate standard operating procedures (SOPs) for checking. The paper focuses on exploring the problem context related to the COVID-19 SOPs for public places. The research on technologies involves using thermal cameras, fingerprint recognition, face recognition, iris recognition, object detection and cloud computing. These technologies can be integrated to provide a more versatile and effective solution. The technological solutions proposed by contemporary researchers are also critically analysed by investigating their advantages and disadvantages. © 2023 IEEE.

PACMAN: A Framework for Pulse Oximeter Digit Detection and Reading in A Low-Resource Setting

In light of the COVID-19 pandemic, patients were required to manually input their daily oxygen saturation (SpO2) and pulse rate (PR) values into a health monitoring system—unfortunately, such a process trend to be an error in typing. Several studies attempted to detect the physiological value from the captured image using optical character recognition (OCR). However, the technology has limited availability with high cost. Thus, this study aimed to propose a novel framework called PACMAN (Pandemic Accelerated Human-Machine Collaboration) with a low-resource deep learning-based computer vision. We compared state-of-the-art object detection algorithms (scaled YOLOv4, YOLOv5, and YOLOR), including the commercial OCR tools for digit recognition on the captured images from the pulse oximeter display. All images were derived from crowdsourced data collection with varying quality and alignment. YOLOv5 was the best-performing model against the given model comparison across all datasets, notably the correctly orientated image dataset. We further improved the model performance with the digits auto-orientation algorithm and applied a clustering algorithm to extract SpO2 and PR values. The accuracy performance of YOLOv5 with the implementations was approximately 81.0-89.5%, which was enhanced compared to without any additional implementation. Accordingly, this study highlighted the completion of the PACMAN framework to detect and read digits in real-world datasets. The proposed framework has been currently integrated into the patient monitoring system utilized by hospitals nationwide. IEEE

Social Distance Monitoring Framework Using YOLO V5 Deep Architecture

Due to the current outburst and speedy spread of the COVID-19 pandemic, there is a need to comply with social distancing rules by the general public. The public needs to, at minimum, hold a distance of 3 ft or 1 m among one another to follow strict social distancing as instructed by using the World Health Organization for general public safety. Researchers have proposed many solutions based on deep learning to reduce the current pandemic, including COVID-19 screening, diagnosis, social distancing monitoring, etc. This work focuses explicitly on social distancing monitoring by a deep learning approach. Here we employ the YOLOV5 object detection technique upon different images and videos to develop a strategy to assist and put strict social distancing in public. The YOLOV5 algorithm is more robust and has a quicker detection pace than its competitors. The suggested object detection framework shows an average precision rating of 94.75%. Some of the existing analyses suffer to identify humans within a range. A few identification blunders happen because of overlapping video frames or humans taking walks too near each other. This detection mistake is due to the overlapping structures, and human beings are standing too close to each other. This paper focuses on correctly identifying humans by using and overcoming the flaws of frame overlapping. Following the proposed social distancing technique also yields positive results in numerous variable eventualities. © 2023, The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd.

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.

COVID-19 Face Mask Classification Using Deep Learning

The COVID-19 pandemic has triggered a global health disaster because its virus is spread mainly through minute respiratory droplets from coughing, sneezing, or prolonged close contact between individuals. Consequently, World Health Organization (WHO) urged wearing face masks in public places such as schools, train stations, hospitals, etc., as a precaution against COVID-19. However, it takes work to monitor people in these places manually. Therefore, an automated facial mask detection system is essential for such enforcement. Nevertheless, face detection systems confront issues, such as the use of accessories that obscure the face region, for example, face masks. Even existing detection systems that depend on facial features struggle to obtain good accuracy. Recent advancements in object detection, based on deep learning (DL) models, have shown good performance in identifying objects in images. This work proposed a DL-based approach to develop a face mask detector model to categorize masked and unmasked faces in images and real-time streaming video. The model is trained and evaluated on two different datasets, which are synthetic and real masked face datasets. Experiments on these two datasets showed that the performance accuracy rate of this model is 99% and 89%, respectively. © 2023, The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd.

Face Mask Detection Using YOLOv3

The COVID-19 is an unprecedented crisis that has resulted in several security issues and large number of casualties. People frequently use masks to protect themselves against the transmission of coronavirus. In view of the fact that specific aspects of the face are obscured, facial identification becomes extremely difficult. During the ongoing coronavirus pandemic, researchers' primary focus has been to come up with suggestions for dealing with the problem through rapid and efficient solutions, as mask detection is required in the current scenario, whether in public or in some institutions such as offices and other workplaces. Only detecting whether a person wears mask or not is not enough. There is another aspect of wearing the mask properly such that it covers all the required portion of the face to ensure there is no exposure to any viruses. To address this, we proposed a reliable technique based on image classification and object localization, which can be accomplished using YOLO v3's object detection in machine learning. © 2023, The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd.

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.

Object detection and activity recognition in video surveillance using neural networks

Purpose: This paper aims to implement and extend the You Only Live Once (YOLO) algorithm for detection of objects and activities. The advantage of YOLO is that it only runs a neural network once to detect the objects in an image, which is why it is powerful and fast. Cameras are found at many different crossroads and locations, but video processing of the feed through an object detection algorithm allows determining and tracking what is captured. Video Surveillance has many applications such as Car Tracking and tracking of people related to crime prevention. This paper provides exhaustive comparison between the existing methods and proposed method. Proposed method is found to have highest object detection accuracy. Design/methodology/approach: The goal of this research is to develop a deep learning framework to automate the task of analyzing video footage through object detection in images. This framework processes video feed or image frames from CCTV, webcam or a DroidCam, which allows the camera in a mobile phone to be used as a webcam for a laptop. The object detection algorithm, with its model trained on a large data set of images, is able to load in each image given as an input, process the image and determine the categories of the matching objects that it finds. As a proof of concept, this research demonstrates the algorithm on images of several different objects. This research implements and extends the YOLO algorithm for detection of objects and activities. The advantage of YOLO is that it only runs a neural network once to detect the objects in an image, which is why it is powerful and fast. Cameras are found at many different crossroads and locations, but video processing of the feed through an object detection algorithm allows determining and tracking what is captured. For video surveillance of traffic cameras, this has many applications, such as car tracking and person tracking for crime prevention. In this research, the implemented algorithm with the proposed methodology is compared against several different prior existing methods in literature. The proposed method was found to have the highest object detection accuracy for object detection and activity recognition, better than other existing methods. Findings: The results indicate that the proposed deep learning–based model can be implemented in real-time for object detection and activity recognition. The added features of car crash detection, fall detection and social distancing detection can be used to implement a real-time video surveillance system that can help save lives and protect people. Such a real-time video surveillance system could be installed at street and traffic cameras and in CCTV systems. When this system would detect a car crash or a fatal human or pedestrian fall with injury, it can be programmed to send automatic messages to the nearest local police, emergency and fire stations. When this system would detect a social distancing violation, it can be programmed to inform the local authorities or sound an alarm with a warning message to alert the public to maintain their distance and avoid spreading their aerosol particles that may cause the spread of viruses, including the COVID-19 virus. Originality/value: This paper proposes an improved and augmented version of the YOLOv3 model that has been extended to perform activity recognition, such as car crash detection, human fall detection and social distancing detection. The proposed model is based on a deep learning convolutional neural network model used to detect objects in images. The model is trained using the widely used and publicly available Common Objects in Context data set. The proposed model, being an extension of YOLO, can be implemented for real-time object and activity recognition. The proposed model had higher accuracies for both large-scale and all-scale object detection. This proposed model also exceeded all the other previous methods that were compared in extending and augmenting the object detection to activity recognition. The proposed model resulted in the highest accuracy for car crash detection, fall detection and social distancing detection. © 2023, Emerald Publishing Limited.

A Comprehensive Alert System Based on Social Distancing for Cautioning People Amidst the COVID-19 Pandemic Using Deep Neural Network Models

The World Health Organization (WHO) has suggested a successful social distancing strategy for reducing the COVID-19 virus spread in public places. All governments and national health bodies have mandated a 2-m physical distance between malls, schools, and congested areas. The existing algorithms proposed and developed for object detection are Simple Online and Real-time Tracking (SORT) and Convolutional Neural Networks (CNN). The YOLOv3 algorithm is used because YOLOv3 is an efficient and powerful real-time object detection algorithm in comparison with several other object detection algorithms. Video surveillance cameras are being used to implement this system. A model will be trained against the most comprehensive datasets, such as the COCO datasets, for this purpose. As a result, high-risk zones, or areas where virus spread is most likely, are identified. This may support authorities in enhancing the setup of a public space according to the precautionary measures to reduce hazardous zones. The developed framework is a comprehensive and precise solution for object detection that can be used in a variety of fields such as autonomous vehicles and human action recognition. © 2023, The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd.

The COVID'19 Dashboard with Implementation of Face Mask Detection and Social Distancing Detection Algorithm

With the outbreak of the global pandemic, India seemed to reach its peak with regard to the number of confirmed positive cases in the months of April and May. Hence, the decision was made to develop a data visualization project with one of the efficient visualization tools Tableau to help people analyze the scenario of the cases across the country. To contribute to state-wise and country-wise analysis of COVID cases in India, 2 dashboards have been developed. The first dashboard consists of the analysis of cases across the country giving a holistic and overall view of the number of deaths, positive cases, and density of cases in each state which is done through color variation. On the other hand, the second dashboard gives a detailed state-wise analysis of cases with the necessary parameters and details catering to every individual state as per the preference of the user. On merging these components, users can get an all-inclusive analysis based on different parameters on the COVID'19 cases across India at a glance. In order to prevent a further spike in cases, implementing a face mask detection system will also take place after conducting a thorough analysis of the possible machine learning algorithms. Two major object detection algorithms were taken into consideration and based on the conclusion drawn, the best algorithm - RCNN was used to implement the face mask detection system. This project is solely motivated by the current extreme situation in the world and as an attempt to provide a solution to combat the same. © 2023 IEEE.

Computer-Aided System for COVID-19 Using Semi-supervised-based Ensemble Learning and Reinforcement Learning

Coronavirus Disease 2019(COVID-19) has shocked the world with its rapid spread and enormous threat to life and has continued up to the present. In this paper, a computer-aided system is proposed to detect infections and predict the disease progression of COVID-19. A high-quality CT scan database labeled with time-stamps and clinicopathologic variables is constructed to provide data support. To our knowledge, it is the only database with time relevance in the community. An object detection model is then trained to annotate infected regions. Using those regions, we detect the infections using a model with semi-supervised-based ensemble learning and predict the disease progression depending on reinforcement learning. We achieve an mAP of 0.92 for object detection. The accuracy for detecting infections is 98.46%, with a sensitivity of 97.68%, a specificity of 99.24%, and an AUC of 0.987. Significantly, the accuracy of predicting disease progression is 90.32% according to the timeline. It is a state-of-the-art result and can be used for clinical usage. © 2022 IEEE.

Mask Recognition through Deep Learning

A major part of computer vision is formed by Object detection. Most of the such tasks are done with efficient object detection. This paper aims to incorporate techniques for facial mask detection to achieve an accurate and efficient mask detection algorithm. The goal is to examine various deep learning algorithms to perform mask detection in this era of Covid. This paper aims on building an application based on facial mask recognition using different deep learning algorithms and compare the results to find out the most accurate algorithm. © 2022 IEEE.

A Review on Recognition of Disguise Face Mask Using Novel Deep Convolutional Neural Network

Object recognition establishes a connection of different objects present in images or videos. Nowadays, this technology is widely used in transportation management systems, intelligence systems, military equipment acquisition, and also in surgical equipment to obtain a surgical guidance, etc. Wearing a facemask has become a mandate in public places to control the spread of coronavirus. This research study has developed a novel facemask detection model based on a single-shot detector (SSD) to collect real-time images. This process has been implemented in three modules: 1) A network of simple error correction features will be introduced based on SSD and partition in order to achieve a better access speed and satisfy the real-time requirements;2) Feature Enhancement Module (FEM) is used to strengthen the in-depth features learned by CNN models to improve the visibility of minor substances;3) A COVID-19-mask will be finally created by considering a large database of face mask images. Test results generate high accuracy while utilizing real-time acquisition and realization of the proposed algorithm. © 2022 IEEE.

MSCAF-Net: A General Framework for Camouflaged Object Detection via Learning Multi-Scale Context-Aware Features

The aim of camouflaged object detection (COD) is to find objects that are hidden in their surrounding environment. Due to the factors like low illumination, occlusion, small size and high similarity to the background, COD is recognized to be a very challenging task. In this paper, we propose a general COD framework, termed as MSCAF-Net, focusing on learning multi-scale context-aware features. To achieve this target, we first adopt the improved Pyramid Vision Transformer (PVTv2) model as the backbone to extract global contextual information at multiple scales. An enhanced receptive field (ERF) module is then designed to refine the features at each scale. Further, a cross-scale feature fusion (CSFF) module is introduced to achieve sufficient interaction of multi-scale information, aiming to enrich the scale diversity of extracted features. In addition, inspired the mechanism of the human visual system, a dense interactive decoder (DID) module is devised to output a rough localization map, which is used to modulate the fused features obtained in the CSFF module for more accurate detection. The effectiveness of our MSCAF-Net is validated on four benchmark datasets. The results show that the proposed method significantly outperforms state-of-the-art (SOTA) COD models by a large margin. Besides, we also investigate the potential of our MSCAF-Net on some other vision tasks that are highly related to COD, such as polyp segmentation, COVID-19 lung infection segmentation, transparent object detection and defect detection. Experimental results demonstrate the high versatility of the proposed MSCAF-Net. The source code and results of our method are available at https://github.com/yuliu316316/MSCAF-COD. IEEE