Social Distancing Detector using Deep Learning

Abstract

Social Distancing is the best possible way to detain the spread of Covid-19. Even though vaccine has been found and working effectively in saving the lives of people, social distancing is necessary to reduce the spread of virus to maximum extent which not only saves people from being infected but also reduces the impact of spreading of the disease. In our proposed system, we use Deep Learning with python to monitor social distancing in public places. This is a software tool that monitor if people are maintaining proper social distancing norms or not by analyzing real time video streams from CC camera. We use YOLO Model which is trained by COCO dataset.

Introduction

I. INTRODUCTION

Covid-19 is the scientific name of corona virus. Till now around 27 crore people were affected by this pandemic including 5 Lakh deaths as per WHO statistics. This disease is considered as a pandemic when it is spread to different countries and caused fatalities. The spread of disease is due to the contact of infected persons with other persons. So, to detain the spread of virus we require an effective monitoring system which monitors people in public places. Monitoring social distancing is very tedious if done manually as it is difficult to monitor continuously by human naked eye. So this is implemented automatically by our software tool which not only monitors people but also highlight the persons who are violating social distancing norms with red color boxes. We use CCTV cameras for automatic monitoring of people.

This toolcan be used in places like Airport, malls etc. The need for developing this tool is to help people and the government to identify and alert people who are being the main cause for the spread of epidemic.Deep learning, a subset of artificial intelligence, has shown remarkable capabilities in various applications, including computer vision. By training neural networks on large datasets, deep learning models can learn intricate patterns and features from visual data, making them well-suited for tasks like object detection and recognition. In the context of social distancing, deep learning algorithms can be trained to detect individuals in crowded scenes and analyze their spatial relationships. This involves identifying people and estimating the distances between them accurately. By employing convolutional neural networks (CNNs) and other deep learning architectures, researchers and developers have made significant strides in creating robust social distancing detection systems.

II. METHODOLOGY

We have used YOLO V3 (Version-3) model which is pre-trained with COCO dataset. We have implemented this project with DNN functionalities like blogFromImage, NMS (Non maxima suppression), dark net. We use blobFromImage functionality for pre-processing of the image, NMS for suppressing the weak signals or weak detections and dark net as a framework of deep neural networks. COCO dataset contains 80 objects but we use only person object out of those objects in the project because we calculate social distancing between persons. If the detected object is person then only we proceed for next steps.

IV. FUTURE ENHANCEMENTS

This tool can be installed in CC cameras for monitoring social distancing in public places like malls, airports etc. We can also Use advance version of YOLO for faster detections in future and we can make GPU as true if we install required packages this will increase speed of execution of output.

Conclusion

Centroid tracking algorithm is used for calculating pairwise distances between the objects. To automate the process of monitoring the social distancing it is an efficient real-time deep learning based framework. The bounding boxes aid in identifying group of people satisfying the closeness property computed using pairwise vectored approach. With Euclidean distance as metric we calculated pairwise centroid distance between detected bounding boxes. The violations are displayed in the output along with violated persons

References

[1] Covid-19 information available: https://covid19.who.int/ [2] Ministry of Health Malaysia (MOHM) Official Portal. [3] COVID-19 (Guidelines), [online] Available: https://www.moh.gov.my/index.php/pages/view/2019-ncov-wuhan guidelines. [4] J. Redmon, S. Divvala, R. Girshick and A. Farhadi, \"You only look once: Unified real-time object detection\", Proceedings of the IEEE conference on computer vision and pattern recognition, pp. 779-788, 2016. [5] K. Simonyan and A. Zisserman, Very deep convolutional networks for large-scale image recognition, 2014. [6] COCO dataset available: https://www.kaggle.com/awsaf49/coco 2017-dataset [7] Harvey A., LaPlace J. 2019. Megapixels: Origins, ethics, and privacy implications of publicly available face recognition image datasets. [Google Scholar] [8] YOLOv3 performance curve available: https://images.app.goo.gl/KSZ8EgaahULzQs2B7 Simonyan K, Zisserman A. Very deep convolutional networks for large-scale image recognition. arXiv preprint arXiv:14091556. 2014 [9] D.T. Nguyen, W. Li and P.O. Ogunbona, \"Human detection from images and videos: A survey\", Pattern Recognition, vol.51,pp. 148-75, 2016. [10] Suresh K, Adwitiya Mukhopadhyay, \"Investigation on COVID-19 by using Machine Learning Techniques\", Computing Power and Communication Technologies (GUCON) 2021 IEEE 4th International Conference on, pp. 1-5, 2021 [11] M.JaithoonBibi, Dr.C.Yamini,(2018) “Survey on classification techniques in data mining”, “International Journal of Theoretical & Applied Sciences”, vol. 10(1): January – June, 2018. 9th February, 2018. ISSN No.:2249-3247 [12] M. Jaithoon Bibi, Dr. C. Yamini,(2019) “An crop yield forecasting using fuzzy logic relationships”, “International Journal of Research and Analytical Reviews”, Pg.no:335-340. vol. 6(1): 1st January, 2019. ISSN No.:2348-1269. [13] Jaithoon Bibi, M., & Karpagavalli, S. (2021). Positional-aware Dual Attention and Topology Fusion GAN for Plant Leaf Disease Image Super-resolution and Classification. Linguistica Antverpiensia, 2021(3),ISSN No.: 0304-2294 Pg.no:(12-25).( Indexed in UGC Care & Scopus) [14] Jaithoon Bibi, M., & Karpagavalli, S. (2021)., \"An Evolutionary Optimization of Positional-Aware Dual-Attention and Topology-Fusion Generative Adversarial Network for Plant Leaf Disease detection\",Turkish Online Journal of Qualitative Inquiry (TOJQI),Volume 12, Issue 3, July 2021:E - ISSN No:1309-6591; Pg no : (2904-2922) ( Scopus, TRDizin) [15] M.Jaithoon Bibi & Karpagavalli.S, A. Kalaivani, (2021), “ Critical Review of Deep Learning Algorithms for Plant Diseases by Leaf Recognition”, Journal of Contemporary Issues in Business and Government”, Pg no: 720 – 729. Vol. 27(5) , 2021. ISSN : 2204-1990; E-ISSN : 1323 – 6903. ( Indexed in UGC CARE List & Web of Science). [16] A. Kalaivani, S. Karpagavalli, M. Jaithoon Bibi (2021), \" A Deep Learning Approach for Real-Time Defect classification in Skin disease\", International Journal of Contemporary Architecture \"The New ARCH\" Vol. 8(2) (2021) ISSN: 2198-7688,PgNo: 443-451.( Indexed in UGC CARE List & Web of Science) [17] M. Jaithoon bibi, Dr. S. Karpagavalli, \" Plant leaf disease image resolution classification using deep convolutional neural network \", International Journal of Natural Sciences, Vol.13, Issue 71, ISSN no: 0976-0997, Pg no40712-40722 [18] Jaithoon Bibi Mohammed Saleem, Dr. S. Karpagavalli, \"Pesticides Recommendation for different leaf diseases and related pests using multi dimensional feature learning deep classifier\", International Information and Engineering Technology Association, Vol.28,Issue 1, ISSN no: 1633-1311 (print); 2116-7125 (online) & https://doi.org/10.18280/isi.280113, Pg.No :133-140 [19] Dr. V. Krishnapriya & M.Jaithoon Bibi,\"\"AN EMPIRICAL ANALYSIS OF METHODS & ALGORITHMS USED IN DETECTION OF SICKLE CELL DISEASE\"\" \"International Journal of Innovative Science and Research Technology\" ,ISSN No:-2456-2165,Volume 9, Issue 1, January – 2024,pg.No:637 to 649 [20] Survey:E-Commerce Comparision App by Figma, \"International Journal of Innovative Science and Research Technology\" ,ISSN No:-2456-2165,Volume 9, Issue 1, January – 2024,pg.No:1284 to 1288

Copyright

Copyright © 2024 Lohith T, Dr. M. Jaithoon Bibi. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.