Classroom AIssistant

front end
front end

Inspiration

With the lack of elementary school teachers in Quebec right now, we wanted to provide a product which makes the teaching experience easier. Our product allows for teachers to focus better on the teaching aspect, and helps identify which students might be struggling to pay attention in class.

What it does

Our program takes a camera feed and finds the faces in it. Then, using our own AI model, it checks whether or not every student is paying attention.

How we built it

We used YOLO7 to recognize the humans in the camera feed/image input. Then, we use Mediapipe to find the faces of the individuals, which our AI model analyze and determine whether or not the individuals are focused.

Challenges we ran into

It was quite complicated to figured out which parameters to train our AI on. It started out being quite inaccurate but after a while we were able to find a model which works.

Accomplishments that we're proud of

Again, we had a lot of trouble designing our model. Coming up with one which has an accuracy of 96% was a very proud moment for us.

What we learned

How to calculate the angles of the head using facial geometry. Additionally, some of us had little to no experience with AI, so that was a learning experience.

What's next for Classroom AIssistant

We'd like to develop a frontend dashboard for the teachers which gives them access to graphs, statistics, so they can better track the behavior of their class.