Automatic Segmentation of an Image

I have downloaded the State Dog Portland Images from Portland University website. I am trying to write an algorithm to automatically segment the dog and person in these images. But I don't know how should I start. I have tried Otsu thresholding and it doesn't work.

Some of the images are attached.

Can anyone guide me to deal with this problem?

Answer 1

[Image Analyst]

Why do you need to do this?

It's a very challenging problem.

The Mathworks had a webinar where they had a cat or dog detector. First they used the Computer Vision System Toolbox to find the animal. Then they used deep learning to identify what kind of animal it was. The webinar is here: Deep Learning Webinar

Good luck.