This is the second demo (first one was rtshapes) of my series of experiments with real time shape recognition using HTML5 getUserMedia, OpenCV and Box2D.
While I was working as a creative technologist at WCRS, I started to play with Box2d a lot. I was also experimenting with Kinect and OpenFrameworks to recognise shapes/blobs in real time, which led me to build some demos mixing both technologies.
In this HTML5 demo, I’m using a webcam to get a live video stream of some shapes drawn in a paper sheet. These video is analysed with JavaScript (using OpenCV) to identify the shapes. Once we have the 2D coordinates of those shapes, I use Box2d to create their representation in the Box2d world.
In the following image, you can see the pieces of paper that are being used to create the pads that will interact with the ball. One pad corresponds to to goalkeeper and the other one to the platform that is used to bounce the ball to try to score a goal.
