Hi everyone, In our measurement course we were asked to represent a new idea using sensors and modern technology.
I really like Deep learning and get inspired by data science. With my friends, we showed a new idea that can help to avoid Covid-19 and keep the production lines on. This was the face mask catcher.
briefly, we found that most workers don't follow instructions in wearing face masks, so we will use the technique of facial recognition in deep learning to catch the one without mask.
First thing, we are going to use a video camera sensor (normal using), which is composed of a sensor called CCD or CMOS. To be honest, I don't know a lot about the sensor's story. All I know is about Deep learning.
Deep learning is a paradigm of Machine learning which uses the idea of the neutral network in the learning process. Machines are like children. They learn by experiments. Like a child asks is that a dog? and you have to answer. Then, he starts storing data.
There is a very famous method used in deep learning, called Viol-Jones Method. It used what is called Haar-feature. In this feature, we want to know if there is something in this area or not. we will consider that there is a nose in this place and we need to know if there is a nose or not. The nose is a feature with three dimensions so it will contain dark areas compared to the face places, which is not contained nose or eyes, etc.
\
I really like Deep learning and get inspired by data science. With my friends, we showed a new idea that can help to avoid Covid-19 and keep the production lines on. This was the face mask catcher.
briefly, we found that most workers don't follow instructions in wearing face masks, so we will use the technique of facial recognition in deep learning to catch the one without mask.
Deep learning is a paradigm of Machine learning which uses the idea of the neutral network in the learning process. Machines are like children. They learn by experiments. Like a child asks is that a dog? and you have to answer. Then, he starts storing data.
\
A square divided into eight squares will cover this area of the face. every square of those eight will take a value between 0 to 1. [0-0.5] brighter than [0.5-1]. Of course, there is no pure 0 or pure 1.
We got the average of the bright squares and black ones. Then, getting the absolute difference between them. If the difference is close to 1, it means there is a feature (nose, hair, eyes, and so on).
What we suggested was so simple, we put two types of data (pictures). people with masks and pictures of workers. If a worker without a mask is found, the camera takes his picture and compares it with the previous data. It gave the name of this worker to the manager.
In conclusion: Our idea is
I'm waiting for your comments.
Comments
Post a Comment