Has anyone in the community tried using the fastai library with the Segment Anything model? If so, I would love to hear about your experience and insights! Alternatively, if you have any ideas on how such models could be utilised for segmentation in the fastai library, I would greatly appreciate your input.
One of the students in my graduate neural networks class has worked on this, but without fastai (I mainly used fast.ai as teaching material). It is probably unnecessary to use fast.ai to use segment anything.
Thank you @msapaydin for the clarity. I have seen people using it for labelling imaging datasets since the model can pick objects in a self-supervised manner and reduce annotation efforts. Thanks again for the clarity.
Thanks @Moody for sharing link to the paper. It is interesting work. This means that SAM needs to be fine tuned on domain-specific datasets such as healthcare and the model can be trained on custom images.