We started with the objective of developing a VR game application for Attention-deficit hyperactivity disorder (ADHD) children. These children do not generally have a long span of attention towards one particular topic. Our initial idea was to develop a VR based game, wherein they could choose the desired object inside the Oculus Rift using a Myo armband. Since we had technical difficulties integrating the Oculus Rift with our laptop, we decided to scale down the project to a gesture-based game.
What does it do
We first developed a single-user Stone-Paper-Scissor game to be played with the computer (which plays randomly). We also designed a web application (launched on Amazon AWS) which provides a runtime video feed of the user input and the randomly generated computer value. The web app then keeps a track of the score. Following this, we then used the user inputs to train a Recurrent Neural Network (RNN) model which learns from this dataset. After a certain pattern is detected on the user inputs, the computer generates sequences with high success likelihood.
Challenges I ran into