A Face Recognition Use Case for Smart Homes “Who are you?”
IoT shapes our life dramatically and finds a lot of solutions for daily problems. If IoT combines with another field such as machine learning, conversational assistance, etc., the number of the idea increases and enables us to think out of the box. In this tutorial, my intention is to give an IoT use-case which can quite helpful for most of us. I called this use-case as “Who are you (WAY)” or it will hinder asking this question whenever someone stays in front of the door and requesting from us to open the door. The traditional way is to ask “WAY”, or open the door or look at through the small door sight, to identify the person. If we replace this system with an AI enabled camera, which is connected to a smart home gateway, it might be not necessary to ask WAY.
How does this system work?
Assuming that someone is outside the home presses the entrance button and there is either only a child or an elderly at the home. She intends to open the door, however, she is not sure whether she can recognize the person staying in front of the door. At this point, the aforementioned system can easily identify the person if there is enough data of the visitor. This brings with itself a privacy issue, since this system enforces the user to share its images with this system, however, it might be the case that the user can share her data just at the moment, where she wants to visit to the home along with the trained model. This approach seems to be not logical, nevertheless it gives users to control when and with which person or institute to share data. Assuming that the user accepts to share her data with the system, which it requests her data along with the trained machine learning model. Now the captured images from the camera can be executed with the Face Recognition Service and the resident can be informed with the recognized person. This simple use case can be today easily implemented with some simple libraries or even most of the cloud platforms offers a number of such libraries.
One can say that she does not want to share her data with the system, however, this would be quite difficult in such an environment, in which we use thousands of devices. Instead of protecting our data and denying sharing it, we may ask of how we protect our data from being tampered. Blockchain technologies give a satisfactory answer, but their deployment will takes still too much time.
In this article, my intention was to share a use case, in which a user recognition service and its daily usage are shown. A more elaborated version of this scenario can be of course designed and implemented. Even though this scenario highlights the user recognition using the image processing, there should be other data types belonging to the humans that represents as a single object on the world or in the universe.
Useful references for face recognition libraries
Top 10 Facial Recognition APIs & Software of 2021 https://rapidapi.com/blog/top-facial-recognition-apis/
https://awesomeopensource.com/projects/face-recognition
A simple face recognition application — https://github.com/ageitgey/face_recognition