In this study, we proposed a method to estimate restaurant congestion and calculate waiting times using live camera footage from the Internet and object detection technology.
Recently, the number of high-turnover restaurants, such as standing dining establishments, has increased, but means to know specific real-time congestion levels or waiting times are limited.
Therefore, in this research, we conducted an experiment to count the number of people inside a store from publicly available live camera footage using the object detection algorithm “YOLOv5”. Specifically, by analyzing the positional relationship between a “reference object” marking the store and the coordinates of detected persons, we determined whether customers were inside the store. This method allows for predicting waiting times based on customer stay duration, and it is expected to enable users to check congestion situations in advance and select restaurants efficiently.
