Crowd Counting during a Pandemic to Find Out Community Response to Activity Restriction Policy Using Deep Learning

ABSTRACT

The pandemic has occurred globally, especially in Indonesia since March 2019. It has been almost 2 years since the danger of spreading the COVID-19 virus was still lurking. Every effort has been made by the government by providing information, counseling, vaccines, and even regulations that limit the level of crowds to a certain scale according to the level of COVID-19 sufferers for each region. Currently, in early 2022, community activities in the city of Malang are allowed to meet face-to-face, such as in the education or office sector. However, if you look at the streets of the Malang city area, you can still see that people are sometimes careless because of the declining death graph due to COVID19. In fact, the government still urges the public to continue implementing the health protocol in various activities and limits the scale of the crowd on certain days. The researcher observes the public's response to all regulations given by the government by detecting objects and crowd counting at two points in the Malang city area, namely Jalan Soekarno Hatta and the entrance to the Malang toll road. The input in the form of photos of street crowds will be processed using computer vision and deep learning to identify the type of object for later analysis of the calculation results according to the calendar and the level of restrictions on current community activities. The percentage of object detection accuracy using deep learning is 80% using a confidence threshold value of 0.3. © 2022 IEEE.