Image Training


I was successfully able to annotate a set of images using prodigy, and I was wondering if there is a way to train the images through prodigy in the same way that I can train NER.


Prodigy doesn't include training recipes for computer vision – there are many different options and frameworks to choose from, so you'd have to decide what you want to use (PyTorch, TensorFlow etc.) and how you want to implement your model. For the NLP tasks, it's easier for us to provide one single training command that calls into spaCy, and it's also easier for us to recommend the best defaults to use. For computer vision, that's a lot less clear, so you'd have to bring your own implementation.

Also see here: