Local OCR
Turn scanned images and PDFs into editable text — right in your browser. Your files never leave your device.
Original file
Recognition result
0 charactersFiles and results stay on this page only. Clear the file or close the page to remove them.
How to use Local OCR
- Drop an image or PDF onto the page, or click “Choose file”.
- Pick the recognition language, rotate the page if it is sideways, then click “Start recognition”.
- Proofread the text in the result box, then copy it or export it as a TXT file.
About this tool
Local OCR runs Tesseract.js — a well-known open-source OCR engine — entirely in your browser, with PDF pages rendered by PDF.js. The engine and the Chinese and English language models (about 8 MB) load once from this site and are then cached by your browser; after that first load, recognition itself needs no network. Your files are never uploaded, and nothing you recognize is sent anywhere.
Frequently asked questions
Are my files uploaded anywhere?
No. Your files stay on your device and recognition runs entirely in your browser. The OCR engine and language models (about 8 MB) load once from our server and are then cached by your browser; after that, recognition itself needs no network connection. Nothing you open here is sent to us or to anyone else.
Which files can I recognize, and what are the limits?
Images in JPG, PNG, WebP or BMP format, and PDF files. One file can be up to 50 MB, and a PDF can have up to 100 pages — longer PDFs need to be split first. Multi-page PDFs are recognized page by page, in order.
Which languages can it recognize?
Chinese and English, mixed in the same document — or Chinese only, or English only, if you pick one in the language menu. Other languages are not supported yet.
What if my PDF is password-protected?
Password-protected PDFs can’t be opened here. Remove the password on your own device first — for example, open the file in a PDF reader and save an unlocked copy — then choose the unlocked file.
How accurate is it, and what should I double-check?
Clean, straight scans of printed text work best; blurry photos, handwriting and decorative fonts are harder. Tables and multi-column layouts may come out in a different reading order than the original, so always check ID numbers, dates and amounts by hand before using them.