Feature Request: Support DeepSeek OCR and other open-weights HuggingFace OCR Models #743
Description
Greetings developers. I rarely use OCR, but have noticed that for now, most OCR tools for open-weights AI are CLI, besides gImageReader, which uses GUI (as I prefer GUIs). It would be nice if gImageReader could support more than just Tesseract, and allow users to download HuggingFace OCR models. Overall, I am looking forward to the future of open-source self-hosted AI.
Here is a list of models that could be listed in an OCR model picker.
OlmOCR: https://huggingface.co/allenai/olmOCR-2-7B-1025
PaddleOCR: https://huggingface.co/unsloth/PaddleOCR-VL
DeepSeek OCR: https://huggingface.co/deepseek-ai/DeepSeek-OCR
DotsOCR: https://huggingface.co/rednote-hilab/dots.ocr
I hope you guys consider this feature request, and thank you for making a great self-hosted OCR tool.
Merry Christmas, and happy holidays
Alex B.