Can someone point me to docs for how to train my own a model?

I would like to build an OCR library and / or a Font detection system. I have a large dataset of images with text I would like to use to train a model.

How do you go about this? I am a C# programmer but also know Python.

Thank you


A good suggestion would be to go through the general HuggingFace tutorial, as this will help you understand the models and how they work:

Once you have the general idea of how the different HuggingFace packages work, you can check which OCR models there are available. Some examples are -

The TrOCR model in the transformers module:

Searching the HuggingFace repos for OCR models:

You can also check in towardsdatascience or in medium to see if there are ocr tutorials using HuggingFace.

1 Like

Thank you. When I get a chance I will read up on this. I wasn’t sure where to start.