| Abstract |
The crowdedness in various places in the city, such as public transportation, restaurants, and public facilities, is high-demand information for not only general people but also municipalities and companies. However, it is not easy to acquire comprehensive data because existing services of crowdedness measurement separately collect and provide data in different ways, although there are many services. This study aims to establish the universal method of crowdedness estimation, which is robust to various environments, by scanning BLE (Bluetooth Low Energy) signals emitted from mobile devices owned by general people. In this paper, we focus on restaurants and public facilities with different types, conditions, and sizes and propose a method of crowdedness estimation by fusing data obtained from other numbers of BLE scanners depending on each space. As a result, we confirmed that models trained with the same feature set for each space show a practical performance. Additionally, we explore the technical challenges when implementing the system in a new space through detailed analysis. |
