OCR language/page segmentation, Google Cloud Vision and FUNSD format

[naourass]

Hello and many thanks for this awesome tool.

For the needs of my current use case, I added these features to my local pawls clone :

• OCR argument for choosing tesseract language and segmentation type
• Google Cloud Vision OCR support which may be more accurate than tesseract in some cases (should download credential file from GCP to root folder)
• Command to generate FUNSD-like json with labels, text and words from coco and tokens csv.

If these features or any of them is relevant to pawls vision, let me know so I can make a PR.

[lolipopshock]

Sounds great -- looking forward to your PR!

[naourass]

Hi there, I've just made a first PR. The second one will be following soon with FUNSD-like and IOB export. Let me know if you have any remarks !