An Optimized Deep Learning Approach for the Prediction of Social Distance Among Individuals in Public Places During Pandemic.

Autor: Sahoo SK; CVR College of Engineering, Hyderabad, Telangana 501510 India., Palai G; Department of Electronics and Communication Engineering, Gandhi Institute for Technological Advancement, Bhubaneswar, Odisha India., Altahan BR; Medical Instrumentation Techniques Engineering Department, Al-Mustaqbal University College, Hilla, Babil 51001 Iraq., Ahammad SH; Department of ECE, Koneru Lakshmaiah Education Foundation, Andhra Pradesh, Vaddeswaram, 522302 India., Priya PP; Department of ECE, Dadi Institute of Engineering and Technology, Anakapalle, Visakhapatnam India., Hossain MA; Institute of Theoretical Electrical Engineering, Faculty of Electrical Engineering and Information Technology, Ruhr University Bochum, 44801 Bochum, Germany.; Department of Electrical and Electronic Engineering, Jashore University of Science and Technology, Jashore, 7408 Bangladesh., Rashed ANZ; Electronics and Electrical Communications Engineering Department, Faculty of Electronic Engineering, Menoufia University, Menouf, 32951 Egypt.
Jazyk: angličtina
Zdroj: New generation computing [New Gener Comput] 2023; Vol. 41 (1), pp. 135-154. Date of Electronic Publication: 2023 Jan 02.
DOI: 10.1007/s00354-022-00202-1
Abstrakt: Social distancing is considered as the most effective prevention techniques for combatting pandemic like Covid-19. It is observed in several places where these norms and conditions have been violated by most of the public though the same has been notified by the local government. Hence, till date, there has been no proper structure for monitoring the loyalty of the social-distancing norms by individuals. This research has proposed an optimized deep learning-based model for predicting social distancing at public places. The proposed research has implemented a customized model using detectron2 and intersection over union (IOU) on the input video objects and predicted the proper social-distancing norms continued by individuals. The extensive trials were conducted with popular state-of-the-art object detection model: regions with convolutional neural networks (RCNN) with detectron2 and fast RCNN, RCNN with TWILIO communication platform, YOLOv3 with TL, fast RCNN with YOLO v4, and fast RCNN with YOLO v2. Among all, the proposed (RCNN with detectron2 and fast RCNN) delivers the efficient performance with precision, mean average precision (mAP), total loss (TL) and training time (TT). The outcomes of the proposed model focused on faster R-CNN for social-distancing norms and detectron2 for identifying the human 'person class' towards estimating and evaluating the violation-threat criteria where the threshold (i.e., 0.75) is calculated. The model attained precision at 98% approximately (97.9%) with 87% recall score where intersection over union (IOU) was at 0.5.
Competing Interests: Conflict of InterestNo competing interests.
(© Ohmsha, Ltd. and Springer Japan KK, part of Springer Nature 2023, corrected publication 2023Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law.)
Databáze: MEDLINE
Externí odkaz:
Nepřihlášeným uživatelům se plný text nezobrazuje