We were talking about common problems we all faced during the pandemic when we all started talking about online school, and we all agreed that we were more distracted during virtual learning than during normal in person instruction. We found it harder to study as we saw progressively more distractions around us. The first step to improving our study skills is to be aware of where they are at so we decided to make a program to give students a score based on how focused they are.

What it does

Our software uses your camera to analyze your face and it will sense when you are distracted and when you are focused. The main features that we added during HackTJ was blink counting and a timer that keeps track of breaks. Since focus directly correlates with how much you blink(less blinks is more focused), we keep track of how much you blink every minute and then update your focus score every minute. The break timer gives you a notification that you can take a break every 25 minutes.

How we built it

We used tensor flow models as the core of our program and we tracked features such as eyelids by using the built-in 3-dimensional vertices system. This basically plots points on your facial features and updates in real-time with your movements. We used JavaScript and HTML5 to write the program and create the website respectively. We also incorporated a timer that manages the students' study time and break/rest time, based on the Pomodoro technique.

Challenges we ran into

We had trouble linking together the JavaScript, HTML, and CSS portions of our app. We aren't very experienced in HTML, CSS, or JavaScript, so we had to learn a lot as we went. It was also somewhat difficult to utilize the TensorFlow model to count blinks.

Accomplishments that we're proud of

We all came into this hackathon with minimal knowledge of JavaScript, HTML5 and CSS, but we were able to use our resources effectively to develop a product we were all proud of.

What we learned

We all learned how to use HTML5 to format webpages and we learned about the tensor flow model and how it tracks joints(in our case facial features).

What's next for DionFocus

We plan to incorporate more details into our program to give a more accurate score such as volume of the room, eyebrow movement, and upper body movement.

We used an opensource TensorFlow facial recognition model from Github to plot the points on the face: link

Built With

Share this project: