Hi I’m trying to do image segmentation.
but I can not understand how it works.
I do know how CNN works in classification (something like MNIST classification) by the way
I just can’t understand how the mask image and labels work in the network.
do I need the datasets having like this???
(image, mask image)
or
[(image, label), (mask image, label)]
having both mask image and label makes me really confusing.
specifically I am trying to do segementation the brain cancer in Brain MRI image
so do I need the dataset like
(brain MRI image, mask image)
or
[(brain MRI, cancer or not),(mask image, cancer or not)]
and another question.
does the visualized weights shows the predicted mask image??