FTC Neural Net Demo

This app demonstrates what a trained neural net can do by detecting FTC game elements and robots. It is fully open-source at http://github.com/kerrj/yoloexampleapp with more extensive documentation of how to use this during the season at http://github.com/kerrj/yoloparser.

To use, enable the camera permission and point the camera at one (or multiple) of the three objects the app is trained to detect:

1. FTC Robot

2. Red wiffle ball (Velocity Vortex)

3. Blue wiffle ball (Velocity Vortex)

The app will draw boxes around detected objects with their name, and percent confidence. The app has a threshold of 20% confidence, so anything below that will not show up.