Our team is excited about robotics and computer vision. We heard about the twitter API at the beginning session and thought it would be unique to integrate twitter into a robotics hack.

What it does

A photo booth that tracks its users and makes sure they are center in frame. Then it can recognize people it has taken pictures of and tweet at them. The photo booth keeps track of who it has already seen and will remember their twitter tags. The program first starts off by finding a person in a frame and rotating the camera so a person is center in the frame. Then a facial detection algorithm takes over and detects if the person is known or unknown. If the person is unknown, we ask them who they are. Once we know who they are, we can tweet and tag them in the picture(@TeamGoose3). This program will remember everyone it has seen.

How I built it

We used a Keysight raspberry pi and a screen along with a servo attached to a Logitech camera. We also used some tape to hold everything together. The software portion uses cv along with a facial recognition library that can both recognize and classify faces. The servo is moved using a special hardware timer servo library.

Challenges I ran into

To access the files on the pi we had to setup a hotspot so we could ssh into the pi. However, it was so slow. So we decided to do all coding off of the pi and then deploy very sparingly.

The camera mount was not that great so we decided for the time being that we would have to elevate the rig for it to work.

Accomplishments that I'm proud of

Implementing face recognition smoothly was very good. Working with so many different aspects of technology and the project still working was amazing. We also made the project look pretty good. It was a success in many different aspects.

What I learned

This was our first time working together as a team, so it really gave us the chance to explore our strengths and get out of our comfort zone. For this project, we had to learn many new skills including Python GUI design, Raspberry Pi OpenCV integration, Twitter API implementation, and how to build a complex computer vision-based project in 24 hours. These are skills that we will be able to carry into the real world, and we can't wait to show them off!

What's next for Gander

More features are coming for Gander! We intend on first implementing a new model that detects side profiles of faces, to better the accuracy of our tracking. Then we want to add multi-face tracking so that you and all your friends are able to share the fun of taking pictures with Gander!

Built With

Share this project: