Inspiration
We wanted to make a machine learning program that can recognize shapes you draw in the air
What it does
It creates a Keras model and allows you to post pictures, but it's a failure since it only works 50% of the time (accuracy ~55%)
How I built it
I used a Keras model with Tensor backend. It's coded in python and was originally intended to take in live video feed and process it at intervals after the shape flashes in game.
Challenges I ran into
Getting the dimensions of images right, it was such a pain.
Accomplishments that I'm proud of
Creating the baseline structure for ML program.
What I learned
Java, Android Studio, Convolutional, Pooling Layers
What's next for balloon powers
A larger project with a more through process
Log in or sign up for Devpost to join the conversation.