Madhav7871/Doc-OCR-Engine ? reverse-engineered prompt

Build me a web app that can upload an Indian identity document image, like an Aadhaar card, and pull out the main details into clean structured JSON. I want a simple React page where someone can drop in a photo, see the extracted name, date of birth, and ID number, and get the results right away.

On the backend, use FastAPI to handle the image upload and run an OCR pipeline that first finds the important fields on the document, then crops those areas, reads the text, and cleans up mistakes so the final output is as accurate as possible. It should be able to handle noisy document photos and different layouts better than basic OCR.

Make the UI straightforward and responsive, and show the extracted data clearly after processing. If you need to check current docs for any libraries, go ahead and look them up online.