For my vision controlled gesture, I created a pong game where the user controls their “paddle” with their movement.<\/p>\n
For easier and more effective tracking I used the color tracking method although point tracking would have worked as well. Because the raw output of the color tracking was very jerky, I filtered and smoothed the output with a running median of the last 5 values. The game also tracks the number of hits and resets with a random starting shot when you miss. For this project I only used a simple AI control that will never miss but this could easily be adapted to fit multiple user inputs to do a player vs player match.<\/p>\n
For the demo I put the video tracking feed as a background to make it easy to visualize the motion.<\/p>\n