A proof of concept has been implemented in 55 classrooms, where we achieved a reliability of approximately 94% in the measurements. Since then, we have used more different types of sensor techniques to find the most cost effective way to measure attendance.

Our experience shows that we spend relatively most of our time at the beginning of any new project, because of the necessary project setup, the privacy requirements and the intelligent customised designs. We also notice that educational spaces can often be very different in terms of location, dimensions, entrance doors, etc., while the counts should preferably work right away. Although we work as standardised as possible, some customisation is usually necessary to get the job done properly and reliably.

As for our own R&D, we at TwiQel developed our own sensor. We can use any other type, no problem, but this one seems promising. It is an advanced device that displays the number of people in almost any room. We have put together a fully-fledged computer with camera technology that, based on a neural network, enables us to count almost real-time. Together with additional A.I., the data is converted into readable information about presence in and use of rooms. The information that is displayed concerns information about attendance per room, per lecture, and per period to be selected. We can even correlate the attendance to the teacher, the classroom, or the weather.

All processing is done on the sensor and no image is ever stored. The sensor delivers only the counts, nothing more. This makes their use accurate and even GDPR- (Dutch: AVG) -proof.  Thanks to the update function via the Twiqel cloud infrastructure, we can ensure that the sensors are permanently updated with the latest software versions. All our customers can benefit from all Twiqel developments. For us, R&D is not a separate cost item; continuous development and continuous improvement are in everything we do.