DeepSOCIAL: Social Distancing Monitoring and Infection Risk Assessment in COVID-19 Pandemic

Abstract

Social distancing is a recommended solution by the World Health Organisation (WHO) to minimise the spread of COVID-19 in public places. The majority of governments and national health authorities have set the 2-meter physical distancing as a mandatory safety measure in shopping centres, schools and other covered areas. In this research, we develop a Deep Neural Network-based Model for automated people detection, tracking, and inter-people distances estimation in the crowd, using common CCTV security cameras. The proposed DNN model along with an inverse perspective mapping technique leads to a very accurate people detection and social distancing monitoring in challenging conditions, including people occlusion, partial visibility, and lighting variations. We also provide an online infection risk assessment scheme by statistical analysis of the Spatio-temporal data from the moving trajectories and the rate of social distancing violations. We identify high-risk zones with the highest possibility of virus spread and infection. This may help authorities to redesign the layout of a public place or to take precaution actions to mitigate high-risk zones. The efficiency of the proposed methodology is evaluated on the Oxford Town Centre dataset, with superior performance in terms of accuracy and speed compared to three state-of-the-art methods.

Deep visual social distancing monitoring to combat COVID-19: A comprehensive survey

“…TSD is a one-video dataset comprising 50 010 frames, which is collected with a 25fps rate and a resolution of 480 × 720. In Rezaei and Azarmi (2020) , Rezaei et al. validate their VSDM system using multi-object annotated datasets, i.e., VOC 2010 ( Everingham et al, 2010 ), COCO, ImageNet ( Russakovsky et al, 2015 ), and Google Open Images (GOI) datasets V6+ ( Kuznetsova et al, 2020 ).…”

“… ( Keniya & Mehendale, 2020 ) CNN, MobileNet-V2 Real-time SD detection Private data Acc=92.8% Validated small image dataset and have moderate performance and privacy issues. ( Rezaei & Azarmi, 2020 ) YOLOv4 Viewpoint-independent pedestrian detection and VSDM VOC, MS-COCO, ImageNet ILSVRC Acc=99.8%) Based on frame-by-frame pedestrian detection. Also, privacy concerns were not addressed.…”

“…Moving forward, the retrieved information is then sent to an IoT sub-network to (i) identify the anonymous IDs of people belonging to a gathering and (ii) provide them with alert messages This framework has been validated on PETS2006 datasets ( PETS2006 database , 2006 ) and other real-world video data recorded from outdoor live cameras in Odessa Mykolaiv ((Ukraine). In Rezaei and Azarmi (2020) , a YOLOv4-based VSDM in the crowd using CCTV cameras is presented, which can be applied either in outdoor or indoor environments. Specifically, an adapted inverse perspective mapping (IPM) approach has been integrated into the VSDM system along with a simple online and real-time tracking (SORT) tracking technique.…”

“…

Fig. 7 Example of the real-time YOLO-based VSDM system in Rezaei and Azarmi (2020) validated on the OTC dataset, which is based on the DarkNet architecture.

…”

“…Additionally, real-time running of VSDM systems can be enabled by implementing them on different types of graphics processing units (GPUs) or powerful central processing units (CPUs). For instance, the solution developed in Rezaei and Azarmi (2020) performs real-time monitoring using either a 10th generation multi-core/multi-thread CPU platform (or higher) or a basic GPU platform. Moving forward, in Fitwi et al (2021) , a powerful Predator Triton 700-A laptop equipped with a GPU card has been used to process more than 20 FPS.…”