How long does class attendance take? 3 minutes? With 60 classes, 4 periods a day, and 180 school days in a year, this program will save a cumulative 72 days every year! Our team recognized that the advent of neural networks yields momentous potential, and one such opportunity is face recognition. We utilized this cutting-edge technology to save time in regards to attendance.

What it does

The program uses facial recognition to determine who enters and exits the room. With this knowledge, we can keep track of everyone who is inside, everyone who is outside, and the unrecognized people that are inside the room. Furthermore, we can display all of this on a front end html application.

How I built it

A camera that is mounted by the door sends a live image feed to Raspberry pi, which then transfers that information to Flask. Flask utilizes neural networks and machine learning to study previous images of faces, and when someone enters the room, the program matches the face to a person in the database. Then, the program stores the attendees in the room, the people that are absent, and the unrecognized people. Finally, the front end program uses html, css, and javascript to display the live video feed, the people that are attending or absent, the faces of all unrecognized people.

Challenges I ran into

When we were using the AWS, we uploaded to the bucket, and that triggered a Lamda. In short, we had too many problematic middle-men, and this was fixed by removing them and communicating directly. Another issue was trying to read from cameras that are not designed for Raspberry pi. Finally, we accidentally pushed the wrong html2 file, causing a huge merge conflict problem.

Accomplishments that I'm proud of

We were successfully able to integrate neural networks with Flask to recognize faces. We were also able to make everything much more efficient than before.

What I learned

We learned that it is often better to directly communicate with the needed software. There is no point in having middlemen unless they have a specific use. Furthermore, we also improved our server creating skills and gained many valuable insights. We also taught a team member how to use GIT and how to program in html.

What's next for Big Brother

We would like to match inputs from external social media sites so that unrecognized attendees could be checked into an event. We also would like to export CSV files that display the attendees, their status, and unrecognized people.

Built With

Share this project: