نظام إنتاجي لاستخراج النصوص من الصور والمستندات الممسوحة بنماذج رؤية محلية بدل OCR التقليدي. يعيد بناء المستند بنفس مواضع النص والجداول، ويعمل على خادم داخلي دون إنترنت، مع ترجمة وأدوات ملفات.
Turn scanned Arabic documents into copyable, layout-faithful text — fully offline.
A production system that extracts text from images and scanned documents using local vision models instead of classic OCR. It rebuilds the page with text and tables in their original positions, runs on an internal server with no internet, and adds translation and file tools.
| Feature | الميزة |
|---|---|
| Accurate Arabic extraction with vision models | استخراج عربي دقيق بنماذج رؤية |
| Layout-faithful replica including tables | نسخة مطابقة للتخطيط مع الجداول |
| Translation and file tools | ترجمة وأدوات ملفات |
| Admin: models, settings, monitoring | لوحة إدارة: النماذج، الإعدادات، المراقبة |
| No login except the admin area | بلا تسجيل دخول إلا للإدارة |
| Layer | Details |
|---|---|
| Stack | PHP · JS (Python fully removed) |
| Inference | Ollama vision models |
| Mode | Offline, internal server |
Note
This repository is a project showcase. The source code is private. For a walkthrough or demo, reach out via it-caf.com.
هذا المستودع للتعريف بالمشروع، والكود المصدري خاص. للعرض التجريبي تواصل عبر it-caf.com.
Built by Abdulrhman Alyazidi · iT CAF — it-caf.com