add non-cv2 crop_image class and call

huggingface / pixparse

Pixel Parsing. A reproduction of OCR-free end-to-end document understanding models with open data

11 stars 3 forks source link

Open molbap opened 11 months ago

molbap commented 11 months ago

Implementation of the suggestion #22 similar to what was already added in transformers' code for Nougat model