Highly Accurate Real time human Counter with Minimum Computation Cost
Shatadal Ghosh, Anurag Kumar, Sriparna Saha · 2021 International Conference on Advances in Electrical, Computing, Communication and Sustainable Technologies (ICAECT) · 2021
During the ongoing Covid-19 Pandemic when we need to operate any public facility like museum, shopping mall, restaurants, or public dealing organizations, we not only need to keep the operations going but also have to ensure precautionary measures to ascertain their safety. As per all SOPs (Standard Operating Procedures) it is advisable to restrict number of visitors inside these enclosed spaces which are most likely to be weather controlled. Automatic safety compliance thus becomes imperative in such situations. Even though absolute compliance and alert signalling will require scrutiny and cross-checking at several levels, a beginning towards automation of compliance monitoring seems mandatory in the neo-normal era. Hence In this project we have designed a low cost rapidly implementable design to monitor the number of visitors inside the self-contained hall. The system will give signal once the maximum permissible visitor population is reached at a given time. Monitoring the optimal population and the density and enforcing visitor to wear mask even within the space manually is tantamount to imposing health hazards to the person who will physically have to monitor and it may as well render the visitors vulnerable. Here we have used Artificial Intelligence based model person detection and tracking. Real time tracking with accuracy is still an important area in computer vision. There are some commercial solution available for the problem but all of them either implemented considering ideal situation or need huge cost and infrastructure. But as a part of museums community we are passing through a financial crises as due to pandemic we closed for visitors. Hence neither we can afford costly system nor a system designed with ideal condition. This motivates us to develop a new system according to our criteria. Here we have modified the available solution for implementation in real world environment using very minimum hardware infrastructure requirements to work on real time with maximum possible efficiency. This system is not only useful for COVID-19 Situation but also its use can be extended beyond the boundary of museums for visitor density monitoring system for large public establishment with minimum computation cost.