How can I make an OCR using python and machine learning libraries?
Unless you are doing it for learning don make one from scratch. Try using Tessaract tesseract-ocr s . It is very good and already has a Deep Learning based model ( tesseract-ocr s ) integrated which works on a lot of data. If your data is too different from how looks in real world try finetuning it (as given here tesseract-ocr s ) . If you are learning some good starting points are CRNN bgshih s and s s . Use PyTorch or Tensorflow to code them up.