Finding relatively large objects in a cluttered background - Help needed

asked 2018-03-14 04:13:31 -0500

questerer gravatar image

I am relatively a newbie to Computer Vision and am struggling with working on complex cluttered images.I need to find relatively large objects that are bigger than a certain threshold in a cluttered environment.I have tried contours with binarizing and also with canny but with no reliable results.The find contours function in opencv doesn't reliably find the contours as desired.Instead of drawing contours around objects, it draws larger contours around the background outside the objects.I have also tried various preprocessing steps but in vain.

Is there any other or better way to detect the larger objects in these conditions? If there is anyone who had solved similar problems like this before, I would love to hear about it and learn as well.

I have included some example images that have cluttered background and large foreground objects in them. It can be anything, a large stand, shopping kart or people that cause obstruction(in any colour). The images that I'm working with have objects in the background that are smaller and cluttered and the foreground object(to be found) is relatively larger. image 1 image 2 image 3 image 4

edit retag flag offensive close merge delete


can you try again with the images ? the lnks are pointing to a website, not to an image

(you could also upload them here, there's an image button in the edit menu)

berak gravatar imageberak ( 2018-03-14 10:03:21 -0500 )edit