Bastian's profile - activity

Does anyone have a good approach to detect certain areas in an image and to receive the coordinates of that area?

I could think of training a neural network, probably a conv net, to detect the area but I'm quite new to opencv and i don't know how to retrieve the area afterwards.

What I actually want to do is to detect the area, retrieve the coordinates and to crop that area. Think of the facebook features of face detection. I want to do something similar but actually not with faces.

I've some knowledge of neural nets, but I can only think of general image classification as a whole but not with segments of the image.