Multiple Edge Computing Devices with Computer Vision for Social Distancing

Coronavirus disease, widely known as COVID-19 is an infectious disease caused by the SARS-CoV-2 virus. Once infected, a person can spread the virus through their nose or mouth in small particles when they cough, sneeze, speak, or breathe. According to the World Health Organization (WHO), one way to be protected from the risk of virus infection is to stay at least 1 meter apart from others while wearing a properly filtered mask. The study aims to design and develop a multiple edge computing system with computer vision capabilities to monitor the adherence of social distancing in multiple locations and in real time. An edge computing device uses a camera to process a stream of images. Graphical Processing Unit (GPU) was utilized for faster inference processing to detect people. The person's location will undergo transformation to get a 2D perspective. Then, a distance calculation algorithm will be imposed to each pair of persons detected to detect breach of social distancing protocol. For every breach detected, location coordinates will be sent to the host database for visualization and monitoring. The use of multiple edge computing devices for computer vision application was compared to the IP camera system in monitoring multiple locations. It is found that utilization of multiple edge computing devices has significant advantages in terms of power consumption, data acquisition, image processing and inference, and setup cost. © 2022 IEEE.

Remote Reading of Surgical Monitor's Physiological Readings: An Image Processing Approach

As a result of the global effect of infectious diseases like COVID-19, remote patient monitoring has become a vital need. Surgical ICU monitors are attached around the clock for patients in critical care. Most ICU monitor systems, on the other hand, lack an output port for transferring data to an auxiliary device for post-processing. Similarly, strapping a slew of wearables to a patient for remote monitoring creates a great deal of discomfort and limits the patient's mobility. Hence, an unique remote monitoring technique for the ICU monitor's physiologically vital readings has been presented, recognizing this need as a research gap. This mechanism has been put to the test in a variety of modes, yielding an overall accuracy of close to 90%.

HUMAN DETECTION AND SOCIAL DISTANCING MEASUREMENT IN A VIDEO

The purpose of this research project is to find the best solution for measuring the distance between people in a video to track the possible COVID-19 social-distancing. This research aims to create a web-application that can be used with closed-circuit televisions (CCTVs) to track positions of persons in interested area and measure distances between any pairs of persons each frame of a video. The process in this project is separated into 3 parts, including 1) tracking positions of people in a video, 2. calibrating camera views, and 3. measuring distances between any two persons. The tracking technique is based on YOLO algorithm, a famous object detection algorithm, that identifies specific objects in the video. In this project, YOLOv3 is used to detect humans to create the bounding box for getting the position in the frame. After getting the bounding box, finding the distance between any pairs in the video is done by using perspective transformation from camera-view into top-down view. Then, the Euclidean distance is used to find the distance of every pair in the video. Any distances closer than 2-meter will be indicated with a line between two people and printed the distance next to the line. The result of perspective transformation is compared with the checkerboard's camera calibration to compare the error rate in several case scenarios. © 2022 IEEE.

COVID-19 Social Distance Surveillance Using Deep Learning

COVID-19 disease discovered from the novel corona virus can spread through close contact with a COVID-19 infected person. One of the measures advised to contain the spread of the virus is to maintain social distancing by minimizing contact between potentially infected individuals and healthy individuals or between population groups with high rates of transmission and population groups with no or low-levels of transmission. Motivated by this practice, we propose a deep learning framework for social distance detection and monitoring using surveillance video that can aid in reducing the impact of COVID-19 pandemic. This work utilizes YOLO, Detectron2 and DETR pre-trained models for detecting humans in a video frame to obtain bounding boxes and their coordinates. Bottom-centre points of the boxes were determined and were then transformed to top-down view for accurate measurement of distances between the detected humans. Based on the depth of each bottom-centre point estimated using monodepth2, dynamic distance between pairs of bounding boxes and corresponding distance threshold (safe distance) to prevent violation of social distancing norm were computed. Bounding boxes which violate the distance threshold were categorized as unsafe. All the experiments were conducted on publicly available Oxford Town Center, PETS2009 and VIRAT dataset. Results showed that Detectron2 with top-down view transformation and distance thresholding using pixel depth estimation outperformed other state-of-the-art models. The major contribution of this work is the estimation and integration of variable depth information in obtaining the distance threshold for evaluating social distances between humans in videos. © 2022, The Author(s), under exclusive license to Springer Nature Switzerland AG.

Social Distancing Crowd Segmentation, Estimation and Visualisation

This research shows a modern crowd counting solution which alters typical prediction solutions into a segmentation of individuals based on a distance threshold, allowing for better visualisation and results. The study proposes using YOLOv4-normal and YOLOv4-tiny models, which have shown great results throughout calibration with an MAE of 14 and 36 respectively. However it did present some issues of accuracy degradation when trained on head annotations at any level of crowd density. As for visualisation, perspective transformation was used which directly helped in providing the distance calculation that was absent from standard transformation. If any variants of YOLOv4 are to be used, the main argument is the choice between speed over accuracy while relying on native implementations. In the case of distance regulation, any transformation that maps itself onto the region of interest, such as perspective transformation should be used to precisely determine distances from a camera to the region of interest itself. © 2021 IEEE.

Design of a Breach Detection System for Social Distancing

The pandemic caused by the 2019 novel coronavirus introduced essential health protocols for everyone's safety. One of which is maintaining a social distance of at least 1 meter as per the guideline set by World Health Organization (WHO). Currently, most spaces were designed prior to the implementation of the social/physical distancing protocol. This project aims to design and develop a detection system utilizing closed-circuit television cameras, to identify spaces where there is a possible breach in the social distancing protocol. The system will generate discrete data to be queried for tabulation, and analysis. The system will also generate a breach map, which indicates the area in the CCTV footage where increasing breaches occur and are marked in increasing color intensity. The system utilized the YOLO V3 object detection algorithm in identifying an object to be human. The system utilized perspective transformation and Euclidean distance estimation in approximating distance for the social distancing protocol. In summary, the human detection accuracy of the system is ≃ 91%, processing at a rate of 30 frames per second in real-time. © 2021 IEEE.