The client wanted to have a precise overview of the number of visitors at their public sports facilities to reduce costs of manually doing the task. In a Proof of Technology, the KPMG NewTech team showed that by using Open Source machine learning models, a very high degree of accuracy was reached.
Off-the-shelf software and hardware can be combined to create advanced and cost-efficient solutions. The hardware prototype was built using Raspberry Pi devices with standard cameras, which were hacked together.
On the software side, the algorithm base was the You Only Look Once open source real-time object detection system. The system is capable of identifying individuals using the public sports facilities to keep track of the usage.
Unlike placing a simple counter at the door, the software allows for further development that will make it possible to identify what specific activities are undertaken by the observed individuals without the need for multiple expensive sensors. This could allow the client to prioritise what kind of equipment to acquire and maintain, to keep their visitors satisfied.