Inspiration
Snapchat is able to detect faces and apply filters real time on a mobile device. We wanted to replicate that in terms of speed and efficiency.
What it does
Using the picamera a live video feed is captured and displayed on the touch screen. The user can then click interactive buttons on the raspberry pi itself to apply a glasses filter and eventually take a picture.
How we built it
We used python to interact with the picamera as well as the GPIO input pins. Additionally we used a python wrapper for OpenCV to get face detection and the filter overlays working.
Challenges we ran into
Opencv needs a strong data set in terms of xml files in order to begin the face detection process, took a while to properly find the ones we needed. Additionally we had to downgrade the camera resolution so that the face detection algorithm could perform faster to simulate real time experiences.
Accomplishments that we're proud of
For relatively little experience with working with the touchscreen and the raspberry pi in general, we were able to successfully have buttons working on a breadboard to interact with our program real time.
What we learned
The facial detection algorithm uses a data set to compare the input matrix images in order to detect faces. Additionally we learned how each of the GPIO pins talk to each other in terms of voltages when configuring the buttons.
What's next for Pi Photobooth
Given more time we can add more filters such as different backgrounds, glasses, and facial gadgets.
Log in or sign up for Devpost to join the conversation.