This project aims to create an electronic people-counting system for large, indoor events at Arizona State University (ASU). Initially, the project will target ASU’s Engineering Career Fair, which must maintain accurate counts of the number of people in each Memorial Union (MU) room to ensure no room exceeds the fire occupancy limit. Currently, ASU employs student volunteers to manually track the number of people in a given room with hand counters. This manual method of counting individuals in a room can be error-prone and inefficient. Hence, this project seeks to increase people-counting accuracy at ASU events, increase the availability of room data to ASU safety officials, and provide a mobile system that can be easily deployed on a per-need basis.
To achieve these three goals, we are developing a device that can be attached to the door frames of event spaces. The device will be independently powered and will not be a permanent fixture so it can be moved based on the owners needs. Each device will contain a Raspberry Pi with a thermal camera. The Raspberry Pi will be loaded with a neural network that can process the images to count each person it registers as entering the vicinity. The count will then be updated and sent to a server to display the live occupancy of each room at the event for ASU staff to monitor. Thus, making the room count more accurate and easier to verify in real time than the current method being employed.