Image segmentation query

Hi I have tried implementing this tutorial shared by you for Image Segmentation using Mask R-CNN. Can you please guide me on how to implement the same algorithm for only one class such as person so that it creates boundaries for only person class in a provided sample? What changes do I need to make to the code provided by you? I am trying to start from scratch. I apologize if this comes off as a stupid question. I am in first year of my Bachelors.

Looking forward to your reply. Thank you.

@mahnoorarshad, This is not a stupid question at all!

Training a custom segmentation model from scratch can be a bit daunting.
The challenging part is to build a suitable dataset - images with appropriate masks(each pixel in the image needs to be assigned to a class - say ‘inside_object’,‘on_border’,'outside_‘object’)
After creating a dataset, the training part should be relatively easy. Use a pre-trained model.

This seems to be a very good guide.
Also refer this

@raj56 and @mahnoorarshad do you know how to determine the parameter of Mask RCNN - VALIDATION_STEPS? Because I saw several examples on the internet with people putting random values in this parameter, but I don’t know how to determine the correct value, would it be from the dataset?

Looking forward to the answer. If you can pass me some material or blog that talks about it, thank you.

© Copyright 2013-2020 Analytics Vidhya