Handwriting OCR
A custom-built optical character recognition model for extracting text from images and documents, including Arabic, English, French, and handwriting.
See what text extraction means
The sample below illustrates the transformation this project is built to support: handwritten document content on one side, readable extracted text on the other.
استمارة طلب تسجيل جديد
يرجى مراجعة البيانات قبل الإرسال
رقم الطلب: ١٢٤٨
اسم المتقدم: يحيى معروف
تاريخ التقديم: ٢١ سبتمبر ٢٠٢٦
Try it with an image
Choose a document image to send it to the OCR service. The extracted text will appear here for review and copying.
Images are sent to the OCR API for processing. Avoid uploading sensitive documents.
1. Add an image
JPG, PNG, or another image format · up to 5 MB
2. Review the result
Extracted content appears in its original reading direction.
Your extracted text will appear here
Problem and purpose
Text embedded in scans, document images, or handwriting is not immediately available as editable text. This project explores extracting that text through an OCR model and provides a browser demo for submitting an image and reviewing the returned result.
How the demo works
- Select or drop an image in the browser interface.
- The image is sent to an OCR API for extraction.
- Review or copy the extracted text from the result panel.
- The demo accepts image uploads up to 5 MB.
The live demo supports Arabic, English, French, and handwritten content, allowing visitors to test the extraction workflow directly with their own document image.
My role and implementation
Yahya Mahroof built the OCR model, the Python processing service, the API integration, and the browser workflow for uploading an image and reviewing the extracted result.