Share:
Privacy: All document processing runs locally in your browser. Documents are never uploaded to any server, keeping your data private and secure.
🖼️ Choose image
Click to choose or drop an image
JPG / PNG / WebP and more supported; clear, high-resolution text works best
Image preview will appear here.
🔤 Recognition progress
0%
Waiting to recognize…
Status: Ready
📄 Recognition result
Technical References
- Tesseract OCR: https://github.com/tesseract-ocr/tesseract
Frequently Asked Questions
Where does the recognition engine run?
The Tesseract.js engine runs locally in your browser — images are never uploaded. The local model must be loaded on first use.
Which languages are supported?
A Simplified Chinese model is provided (which also recognizes English), and results can be exported as TXT.
Why is the first recognition slow?
The first run loads an ~20MB local model. Precise mode runs multiple enhanced recognition passes, so it takes longer than fast mode.
How can I improve recognition accuracy?
Use 'Precise Mode': it enhances contrast, binarizes, and enlarges the image, then runs multiple passes to pick the best result. For screenshots, set the layout type to 'Sparse Text'.
What affects recognition quality?
Sharper images and higher text-to-background contrast produce better accuracy. Handwriting and heavily skewed photos work less well — crop or rotate them first.
Technical References
- Tesseract OCR: https://github.com/tesseract-ocr/tesseract