I am trying to adapt the EAST code to use with the fastai library. You can find the source code here. This problem finds the boxes or where there is text, regardless of the text context. In other words, it only does regression, not classification.
I see that in v1 the Base class is DatasetBase. But even that class has
c as the number of classes. How should I go for making a custom dataset for regression on images? I have found some good resources but only to be applied in v.0.7, and I need v1.0.x optimizers.