Distraction_Reminder

logo
So proud to get 80% accuracy!
Program Test
Before taking the Video

Distraction Alarmer

This program uses a trained CNN model to detect if the one in front of the camera is distracted. We used roughly 1500 images of focused and distracted eyes to train the deep learning model, and the accuracy we got was about 80%. It's absolutely a wonderful model.

In the folder, there are two .py files. The main.py opens up the camera, take a shot, and do the prediction. The train.py generates a deep learning model. The private images we used to get this model is not submitted.

This program is Charlie Guo, Edward Li, Tracy Lin's Hackathon project. It is aimed to help students under the Covid-19 pandemic to focus in online classes (we wish it could work:). Moreover, it can be used in a car to remind drivers when they are distracted.

alt text