🩴 Add pre-process step to correct image orientation

getomni-ai / zerox

PDF to Markdown with vision models

https://getomni.ai/ocr-demo

MIT License

6.74k stars 367 forks source link

🩴 Add pre-process step to correct image orientation #52

Closed tylermaran closed 1 month ago

tylermaran commented 1 month ago

Zerox performs pretty poorly when the image is sideways or upside down. Add a preprocess step to check text orientation and correct before passing to gpt

tylermaran commented 1 month ago

Example of running the flip:

https://github.com/user-attachments/assets/32673656-ae10-4c44-892b-85e071c23308