Social Distancing Violation Detection in Video Using ChessBoard and Bird's-eye Perspective

Thousands of infections, hundreds of deaths every day - these are numbers that speak the current serious status, numbers that each of us is no longer unfamiliar with in the current context, the context of the raging epidemic - Coronavirus disease epidemic. Therefore, we need solutions and technologies to fight the epidemic promptly and quickly to prevent or reduce the effect of the epidemic. Numerous studies have warned that if we contact an infected person within a distance of fewer than two meters, it can be considered a high risk of infecting Coronavirus. To detect a contact distance shorter than two meters and provides warnings to violations in monitoring systems based on a camera, we present an approach to solving two problems, including detecting objects - here are humans and calculating the distance between objects using Chessboard and bird's eye perspective. We have leveraged the pre-trained InceptionV2 model, a famous convolutional neural network for object detection, to detect people in the video. Also, we propose to use a perspective transformation algorithm for the distance calculation converting pixels from the camera perspective to a bird's eye view. Then, we choose the minimum distance from the distance in the determined field to the distance in pixels and calculate the distance violation based on the bird's eye view, with camera calibration and minimum distance selection process based on field distance. The proposed method is tested in some scenarios to provide warnings of social distancing violations. The work is expected to generate a safe area providing warnings to protect employees in administrative environments with a high risk of contacting numerous people. © 2022, The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd.

K-Zones: a Machine Learning-Based System to Estimate Social Distancing Violations During Pandemic Eras

The outbreak of pandemics adversely influences various aspects of people's lives, including economies, education, careers, and social relations. Therefore, many authorities worldwide resort to imposing social distancing regulations to flatten the curve of new confirmed cases. This paper proposes a Machine Learning-based social distancing violation detection system. Unlike many contributions in the literature that use pairwise distance computation running in quadratic execution time, this paper introduces a novel technique that runs in linear time. The solution is considered a Video Surveillance System, and the experimental results show how the system effectively detects not only social distancing violations but also the severity of those violations. © 2022 IEEE.