[Beginner] Recognizing a logo

asked 2018-05-07 19:45:10 -0500

I am building a 'follow bot' and wanted to use a logo as the target for the bot to follow. Essentially the logo will be on a button that I will stick on my back. The bot runs a Raspberry Pi with OpenCV2 and tensorflow on it and everything is written in Python so far. At this point everything else works but I'm not sure what approach I should use when training the logo for OpenCV.

The logo has bitmap and vector options available to me but not sure that really matters.

What training method should I use? Any suggestions for approach or sample code for navigation based on object detection?


Is training a logo the same as training anything else for object recognition and I should jsut use the basics?


edit retag flag offensive close merge delete



you probably do not need to train your own markers, but instead simply use the default aruco markers embedded in OpenCV

berak gravatar imageberak ( 2018-05-07 21:00:19 -0500 )edit