MohamedRashad/Dots-OCR

[Admin maintenance] Support new ZeroGPU hardware (#4)

Enhance is_arabic_text function: refine detection to focus on headers and paragraphs, excluding lists and code blocks

Enhance layoutjson2md function: embed images in markdown output using base64 encoding for 'Picture' category

Refactor turn_page function: update return values to include processed image and layout JSON, and improve page info handling

Refactor create_gradio_interface: remove unused processing state and clean up return values

Add Arabic text detection functionality and update UI for multilingual support

Remove debug print statement for torch version from app.py

Refactor layoutjson2md function: remove unused parameter and clean up markdown output handling

Update inference decorator to use GPU() from spaces for improved performance

Refactor inference and process_image functions: streamline parameters and enhance layout processing

Update transformers version in requirements.txt to 4.51.3

Remove 'accelerate' from requirements.txt and add 'spaces' dependency

Refactor app.py and requirements.txt: Reorganize imports and add 'accelerate' to dependencies

Update flash-attention wheel URL in requirements.txt

Refactor app.py: Clean up unused prompts and reorganize imports for clarity

Remove flash-attention prebuilt wheels URL from requirements.txt

Remove transformers GitHub repository URL from requirements.txt

Add Dots.OCR GitHub repository URL to requirements.txt

Update transformers dependency to use GitHub repository URL in requirements.txt

Add torchvision to requirements.txt

Add initial implementation of Dots.OCR Gradio demo application and requirements

initial commit