detect object of interest in an image using opencv

asked 2017-03-08 00:49:48 -0500

This question might come across as really generic, since I dont have much experience with opencv, so please pardon me. I am looking for a way to detect objects of interest in an image (not just faces, but any foreground) and get a bounding rectangle for all possible objects of interest in the image. Is there a way to do so using opencv? What modules should I be looking at? So far in my research, I have tried using Saliency API of opencv and some of the feature detection algorithms. They mostly work but they dont really pin point the bounding area of the object. Can I use its output and use some other algorithm to process that data and get a bounding rectangle of possible objects of interest?

Thanks

edit retag flag offensive close merge delete

Comments

1

"object segmentation" is an active research topic, don't really expect readymades here.

sure, you can use findContours(), and get the bounding box for the contour, that best fits your found region.

also have a look at cnn's e.g. https://pjreddie.com/media/files/pape...

berak gravatar imageberak ( 2017-03-08 01:10:45 -0500 )edit