Real-Time Social Distancing Detector Using Socialdistancingnet-19 Deep Learning Network

Keniya, Rinkal; Mehendale, Ninad

doi:10.2139/ssrn.3669311

Cited by 23 publications

(8 citation statements)

References 0 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…There is also a method for detecting close-contacts based on the size of the overlapping area of the bounding boxes [3]. However, since the distance between people is not calculated, we cannot compare our system with it.…”

Section: ) Localization Comparison With Other Methodsmentioning

confidence: 99%

“…However, if a camera captures only the upper body, the bottom edge of the bounding box corresponds to the waist position, leading to position error. Additionally, M. Rezaei et al [2] and R. Keniya et al [3] did not calculate the actual interpersonal distance but estimated whether each person violates social distance based on the size of the bounding box. However, the size of the bounding box varies significantly depending on the pose.…”

Section: B Infection Preventionmentioning

confidence: 99%

See 1 more Smart Citation

Close-Contact Detection Using a Single Camera for Sports Considering Occlusion

et al. 2022

View full text Add to dashboard Cite

The Coronavirus disease 2019 (COVID-19) is still prevalent in the world. Exercise is important to maintain our health while dealing with infectious diseases. Social distancing is more important during exercise because we may not be able to wear masks to avoid breathing problems, heatstroke, etc. To maintain social distancing during exercise, we develop a close-contact detection system using a single camera especially for sports in schools and gyms. We rely on a single camera because of the deployment cost. The system recognizes people from a video and estimates the interpersonal distance for close-contact detection. The challenge is the occlusion of people, which leads to false negatives in close-contact detection.To solve the problem, we leverage the observation that most false negatives in human detection are caused by occlusion owing to other people. This is because there are few obstacles in sports facilities. Based on the above observation, we assume that a person still exists near the last detected position even when s/he disappears in the proximity of other people. For evaluation, we recorded 834 videos that were 112 min long in total including various scenarios with 2724 close-contacts. The results show that the F1-score of closecontact detection and tracking are 83.6% and 67.3%, respectively. We also confirmed that the start and end time errors are within 1 s for more than 80% of the close-contacts.

show abstract

Section: ) Localization Comparison With Other Methodsmentioning

confidence: 99%

Section: B Infection Preventionmentioning

confidence: 99%

Close-Contact Detection Using a Single Camera for Sports Considering Occlusion

et al. 2022

View full text Add to dashboard Cite

show abstract

“…The paper introduces a distance calculation algorithm to calculate the social distance score of a segmented image using HR-Net segmentation model. The core segmentation model is developed using HRNET and object contextual representation transformer architecture [43]. The transformer pipeline is illustrated in Fig.…”

Section: Social Distance Calculationmentioning

confidence: 99%

HRNET: AI-on-Edge for Mask Detection and Social Distancing Calculation

Sengupta

Srivastava

2022

SN COMPUT. SCI.

View full text Add to dashboard Cite

The purpose of the paper is to provide innovative emerging technology framework for community to combat epidemic situations. The paper proposes a unique outbreak response system framework based on artificial intelligence and edge computing for citizen centric services to help track and trace people eluding safety policies like mask detection and social distancing measure in public or workplace setup. The framework further provides implementation guideline in industrial setup as well for governance and contact tracing tasks. The adoption will thus lead in smart city planning and development focusing on citizen health systems contributing to improved quality of life. The conceptual framework presented is validated through quantitative data analysis via secondary data collection from researcher's public websites, GitHub repositories and renowned journals and further benchmarking were conducted for experimental results in Microsoft Azure cloud environment. The study includes selective AI models for benchmark analysis and were assessed on performance and accuracy in edge computing environment for large-scale societal setup. Overall YOLO model outperforms in object detection task and is faster enough for mask detection and HRNetV2 outperform semantic segmentation problem applied to solve social distancing task in AI-Edge inferencing environmental setup. The paper proposes new Edge-AI algorithm for building technology-oriented solutions for detecting mask in human movement and social distance. The paper enriches the technological advancement in artificial intelligence and edge computing applied to problems in society and healthcare systems. The framework further equips government agency, system providers to design and construct technology-oriented models in community setup to increase the quality of life using emerging technologies into smart urban environments. Keywords Deep learning • Edge-computing • Social and industrial safetyThis article is part of the topical collection "Advances in Computational Approaches for Artificial Intelligence, Image Processing, IoT and Cloud Applications" guest edited by Bhanu Prakash K. N. and M. Shivakumar.

show abstract

“…In this scenario the demonstrated tools fulfils a practical need, dynamically monitoring the area and making possible the extension of the infrastructure using a data-driven approach. The recent shock caused by the COVID pandemic impacted also the airport areas, so the social distancing enforcement has emerged as an additional requirement to be enforced, so dynamic detection of people clusters (as in Figure 10) and avoiding people clustering [22], sticking in the limits imposed by law, has come to be very important. It has been easily determined that the position tracking could drive suggestions in this perspective, so if a shop has reached the maximum allowed number of customers, the traveller could be advised about less busy alternatives, and some virtual queues could be setup to alert interested people when space is available and it is their turn.…”

Section: Figure 9 Managing Multiple Fogs In a Smart City Scenariomentioning

confidence: 99%

Towards the Internet of Behaviors in Smart Cities through a Fog-To-Cloud Approach

Salis¹

2021

HighTech. Innov. J.

View full text Add to dashboard Cite

Recent advances in Internet of Things (IoT) and the rising of the Internet of Behavior (IoB) have made it possible to develop real-time improved traveler assistance tools for mobile phones, assisted by cloud-based machine learning, and using fog computing in between IoT and the Cloud. Within the Horizon2020-funded mF2C project an Android app has been developed exploiting the proximity marketing concept and covers the essential path through the airport onto the flight, from the least busy security queue through to the time to walk to gate, gate changes, and other obstacles that airports tend to entertain travelers with. It gives chance to travelers to discover the facilities of the airport, aided by a recommender system using machine learning, that can make recommendations and offer voucher according with the traveler’s preferences or on similarities to other travelers. The system provides obvious benefits to the airport planners, not only people tracking in the shops area, but also aggregated and anonymized view, like heat maps that can highlight bottlenecks in the infrastructure, or suggest situations that require intervention, such as emergencies. With the emerging of the COVID pandemic the tool could be adapted to help in the social distancing to guarantee safety. The use of the fog-to-cloud platform and the fulfilling of all centricity and privacy requirements of the IoB give evidence of the impact of the solution. Doi: 10.28991/HIJ-2021-02-04-01 Full Text: PDF

show abstract

Real-Time Social Distancing Detector Using Socialdistancingnet-19 Deep Learning Network

Cited by 23 publications

References 0 publications

Close-Contact Detection Using a Single Camera for Sports Considering Occlusion

Close-Contact Detection Using a Single Camera for Sports Considering Occlusion

HRNET: AI-on-Edge for Mask Detection and Social Distancing Calculation

Towards the Internet of Behaviors in Smart Cities through a Fog-To-Cloud Approach

Contact Info

Product

Resources

About