Object Detection with Python, Best Method?

asked 2013-11-18 09:08:27 -0500

cwalke gravatar image

updated 2013-11-18 09:53:20 -0500

berak gravatar image

I would like to use OpenCV to make a program that detects objects in video. The camera would point up at a clear/transparent background and would see different objects, most likely up to 8 and trigger an action based on what object it detects. I can make the objects have a customized pattern that is associated with that object. Multiple objects will be placed on the surface at a time, meaning multiple objects will need to be detected at the same time.

What is the best method to accomplish this? Should I use Haar Cascades and train each unique object pattern or would template matching work just as reliably? It seems custom learning each object would be time consuming so I want to make sure it is the best option before I go down that road.

edit retag flag offensive close merge delete