Hi, I'm new top computer vision and need some advice where to start. I've written a application which displays a Rubik cube and lets me manipulate the cube like a real one. I also integrated a auto-solver button. Then I thought it would be nice if that button also existed in real life. So, I want to create an application which analyzes the cube with a camera and solves the cube with a robot. The first step would be to detect and analyze the cube through a camera. For this I think I need 3 things: - detect if the cube is visible by the camera - find out where in the picture the cube is and find the state of the current face (the colors of the sub-faces) - some kind of motion tracking to stitch the detected faces together. I went through the tutorials on opencv.org and was googling about feature detection / description / matching and object detection with haar / lbp. But I'm still not sure what's the best option to start with. - are features good enough - should I go for LBP - or even something completely different like neural networks
Can you please tell me how you would detect and analyze a Rubik cube? Thanks
Regards Bernhard