| --- |
| title: Arabic PDF OCR |
| emoji: 🕌 |
| colorFrom: green |
| colorTo: blue |
| sdk: gradio |
| sdk_version: 4.44.0 |
| app_file: app.py |
| pinned: false |
| license: mit |
| --- |
| |
| # 🕌 Arabic PDF OCR |
|
|
| Extract text from Arabic PDFs using open-source OCR engines built for Arabic. |
|
|
| | Model | Quality | Notes | |
| |---|---|---| |
| | **QARI-OCR** | ⭐⭐⭐⭐⭐ | Built specifically for Arabic by the NAMAA community. Handles diacritics, mixed fonts, complex layouts. | |
| | **EasyOCR** | ⭐⭐⭐⭐ | Deep-learning, great balance of speed and accuracy | |
| | **PaddleOCR** | ⭐⭐⭐⭐ | Excellent for clean printed Arabic text, very fast | |
|
|
| > For best results, enable a GPU in Space settings (ZeroGPU or T4). |
|
|
