Social distance monitoring framework for Covid-19 using deep learning

• Main Authors: Gogulamudi, Lakshmi, Mudigonda, Supraja, Renduchintala, Yagna Priya, Vemula, Bhargavi, Palnati, Gowtham Harsha
• Format: Conference Proceeding
• Language: English
• Subjects: Algorithms; Closed circuit television; Deep learning; Machine learning; Monitoring; Social distancing; Violations
• ISSN: 0094-243X; 1551-7616
• DOI: 10.1063/5.0148967

These days, social distancing measures are very vital to dropping COVID spread. In order to break the chain of the spreading of this virus, social distancing is sternly followed as a standard. The paper validates a system that would be useful in strictly monitoring public places such as the ATMs, malls, and the hospitals and passport offices for any social distancing violations that takes place. So with the benefit of this anticipated system, it would be expediently likely to monitor personalities, whether they are sustaining social distancing in the zone under observation, and also, to aware the people, when there were any defilements from the predefined limits. This deep learning system could be used for exposure within asure limited reserve. This algorithm can be executed on the live pictures of CCTV photographic camera to perform the required task. This simulated model uses the basic deep learning algorithms with Open CV library for evaluation the distance amongst the people within frame. Also, a yellow model which is trained on the COCO dataset to identify the people within the frame. The system has to be configured according to the position it is being installed. By executing this algorithm, the count of violations is informed based on the distance and the set threshold number of violations reported are 1 and 2 for real-time images correspondingly. The red colour bounding boxes highlighting the violations are displayed on the screen itself. Laterally, with the distance reporting, effectiveness and correctness were authenticated for a greater number ofsamples.