Over the past few weeks, it has become clear that we cannot put our masks into storage just yet. Some people are even predicting that they will become a permanent part of our outfit. We all now know that wearing a mask is an important shield in the fight against the coronavirus. However, in practice there are still people who do not wear them or at least not properly. We trained an algorithm to detect this. By installing a smart camera, you can send targeted passers-by or customers a message to say that a mask is mandatory or even deny them access if they are not wearing a mask.
Many people don't like telling strangers what to do. If you point out to someone that they need to wear a mask over their nose too to be effective, you'll quickly be labelled as a know-it-all or busybody. By using technology to convey this message, we remove the personal charge and nobody slips through the cracks.
The technology used here is Artificial Intelligence and Machine Learning. All building blocks are present, with a camera, a 4G modem, an edge computing device on which the algorithm runs and a cloud platform for capturing and interpreting the data. For the more technical among us, you might also like know that the model was trained using the yolo model (YoloV4) and Darknet using transfer learning techniques. It was therefore developed using fully open source technologies in Python, so you are not entirely dependent on a cloud provider. A fairly simple construction in itself, but one that can make a big difference in the fight against the virus.
There are many areas of application where this piece of intelligence could prove useful, e.g. an automatic message in bus or tram when more than 5% of passengers aren't wearing their masks or at the entrance to shopping malls. It would also fit in perfectly with the bigger picture of a Smart City. After all, we use new technologies to make cities safer by turning raw data into smart insights. In addition to detecting masks, the same set-up could also be used to see how people move through a city or to count passers-by, monitoring and encouraging social distancing in real time. All data is centralized on a data platform, fully compliant with GDPR rules, of course. The idea is to make data usable by giving it the right meaning, correlating different data sources and building intelligent applications. The ultimate goal is to improve urban life or organize the city better. Based on the data, for example, it is possible to make a dashboard which shows you when a shopping street is busy. A barometer like this helps to prevent crowds, which is quite important in these times.
