Most European governments decided to make it mandatory to wear a face mask in public transport. In many other industries face masks protect the health of employees. Is it necessary to enforce these regulations, or will people voluntarily wear aerosol protection out of respect for the safety and health of eachother? How can we monitor or maintain these new rules and behavioral changes? And how to help security officers doing their jobs?
The answer can be found in applying artificial intelligence. Automated enforcement or computerized reminders through artificial intelligence, take the pressure off staff and can give clues and insights into people’s behavior in a neutral way, without the violation of privacy regulations. As a manufacturer of AI camera’s, AVUTEC believes computerized intelligence can be a wonderful digital assistant.
The rapid development of processor technology lead to hardware boosted neural processing power. This paved the way for smart devices operating on the edge. It inspired AVUTEC to develop the Gatekeeper X, an AI powered visual sensor. A cameras that can multitask, perform complex video analysis and communicate results to third party systems simultaneously, without the need for extra server capacity.
The development of lean machine learning and deep learning algorithms added speed and functionality, optimizing hardware usage. As all AVUTEC devices operate on the edge, recognition and detection occur inside of the sensor. People are not identified, only detected, guarding privacy.
Face mask detection is the combination of two deep learning networks. The first network detects all the faces in an image. The faces are cut out and fed to the second network to determine the presence of a face mask. After all faces are analyzed, the image is put back together. A neural net can be compared to a tiny part of a human brain. It is a network of neurons trained for a specific task. It would take a magnitude of interconnected networks for devices, to be able to multitask just like us. The Gatekeeper X can run three to four networks in parallel, which is quite an achievement for an embedded AI camera!
Face mask detection
The deployment of face mask detection can be done in successive stages of interference. Important to bare in mind is that AVUTEC face mask detection is done anonymously on the edge, no ethernet is needed, no exchange of data is required. Let’s take a closer look.
The overall detection of face masks with the help of technology in public places, generates insights about human behavior. Analysis of the images has been done inside the camera, saving an enormous amount of handcrafted work. The data generated provides a solid understanding whether people wear their face masks or not.
A gentle reminder
Placing a camera at the entrance of a railroad station takes things a step further. The sensor can be programmed to remind passengers to wear their masks by displaying messages on a matrix board. The same way signs along the road inform us about our actual traffic speed. Instead of “You are speeding” the message could be “Don’t forget to wear your mask”.
Again a step ahead is a camera instructed to only allow people in when wearing face protection, just like on the airport. In this scenario the detection is coupled to an access control unit and only opens for a person wearing face protection. The unit is not coupled to ethernet nor to a personalized public transport card, protecting privacy.
As a society we decide the level of artificial intelligence we believe is acceptable. The European law is strict about privacy which is a valuable good. Like all technologies it can be used righteous or unjust. In the last decades smart technology has become our companion and we discovered the fantastic possibilities it offers. As artificial intelligence will evolve, it is important to constantly redefine the role this technology will play in our lives.