facebookresearch / nougat

Implementation of Nougat Neural Optical Understanding for Academic Documents
https://facebookresearch.github.io/nougat/
MIT License
8.81k stars 561 forks source link

Can nougat handle images e.g. .jpeg, .png that might have maths? #146

Open brando90 opened 11 months ago

brando90 commented 11 months ago

or is it better to convert those to pdf then have nougat process them?

smclh commented 11 months ago

image

MereHappyJaZZ commented 11 months ago

@smclh Where is that notebook from? Anything you can share?

wufan-tb commented 11 months ago

@smclh Hi,I tried your code, yet I got a warning and an error, have you seen these in your expriments? image image

smclh commented 11 months ago

v0.1.17 image

wufan-tb commented 11 months ago

I found my nougat version is 0.1.17 too, could you please tell me you python and pytorch version? image

smclh commented 11 months ago

torch 1.10.0+cu113 python 3.8.8

wufan-tb commented 11 months ago

@smclh Hi, I tried another conda environment(with python 3.9 and torch 1.12.1), this time it works, thank you very much! image

luckymore commented 5 months ago

@MereHappyJaZZ https://github.com/facebookresearch/nougat/blob/notebook-demo/notebooks/nougat.ipynb