Detecting Social Distancing Violation Using Convolutional Neural Network

Detecting Social Distancing Violation Using Convolutional Neural Network

                    Makrand Bhonde1, Rutwik More2, Kunal Wade3, Harshal Suple4, Dr. K. P. Wagh5

                   Department of Information Technology, Government College of Engineering Amravati
Kathora Naka, Amravati, India 444602

Abstract— The paper describes an approach to detect violation of social distance between people. People are identified using a special object detection model YOLO which is based on Convolutional Neural Network and a pre-trained file yolo-weight. After the input video file and detecting human objects the distance between them is calculated by camera view calibration and coordinates assigned by object detection. The output video file contains the results where if the social distance is violated then it encounters a red frame, otherwise for safe distance it is represented by a green frame. As per standard guidelines the social distance is kept to be 6 feet or 2 metres. This model is tested on a pre-recorded video but it has a potential future application where it can be integrated to a live surveillance for real time detection.

Keywords— social distance, YOLO, convolutional neural network, object detection