Pulling text out of a screenshot
Paste a chat screenshot or an error dialog with Ctrl+V, recognize it locally and copy the text; the image never leaves your computer.
Loading…
Guide
Dual-engine OCR turns the text inside an image into editable text and can run two engines at once, placing their results side by side: the local engine is Tesseract (Simplified Chinese + English, with the engine and language data self-hosted on this site, so the image never leaves the browser), and the cloud engine is OCR.space (language=chs, OCREngine=2, with the image uploaded to api.ocr.space).
Updated 2026-09-093 min read
Dual-engine OCR turns the text inside an image into editable text and can run two engines at once, placing their results side by side: the local engine is Tesseract (Simplified Chinese + English, with the engine and language data self-hosted on this site, so the image never leaves the browser), and the cloud engine is OCR.space (language=chs, OCREngine=2, with the image uploaded to api.ocr.space).
If you only need offline recognition, the local engine is enough. When tables or complex layouts come out poorly, add a free OCR.space key and switch to the cloud engine or run Compare directly. For batch recognition use the OCR tool in the Documents category; this one handles a single image at a time.
| Input | Output | Notes |
|---|---|---|
A screenshot of Chinese meeting notes, local engine |
Recognized text with an “87% confidence” badge |
The confidence is Tesseract's overall average |
The same image with Compare selected |
Left: local engine (Tesseract); right: OCR.space (cloud) · Engine 2 · Chinese Simplified |
The two run independently, so a failure on one side does not affect the other |
language=chs.Paste a chat screenshot or an error dialog with Ctrl+V, recognize it locally and copy the text; the image never leaves your computer.
When a scanned contract produces typos locally, add an OCR.space key, choose Compare and stitch together the more accurate passage from each side.
The local engine is cached after its first load, so Chinese and English recognition keeps working on a poor connection.
On first use it loads the Tesseract engine and the Chinese/English language data from this site (a few megabytes, self-hosted rather than served from a third-party CDN) and reuses them from the browser cache afterwards. Images are never uploaded.
The whole image file, the language parameters and your API key, POSTed as a multipart form to api.ocr.space/parse/image. Do not use the cloud engine for images that contain ID cards or bank cards.
Register for free at ocr.space/ocrapi/freekey, then enter it under Settings → API keys in the OCR.space field; the key is kept in this session's memory only by default.
With Local selected the image never leaves the browser; only the recognition engine and language data are loaded from this site on first use. With Cloud or Compare selected, the image and your OCR.space key are sent to api.ocr.space, which this site neither handles nor stores. Use the local engine only for images that contain sensitive information.
Updated 2026-09-09
Local Tesseract and cloud OCR.space side by side; for batches use the local OCR under Documents
The local engine is active, so the image never leaves the browser
· Local engine: Tesseract for Chinese and English, fully offline and free; language packs are self-hosted with the site
· Cloud engine: OCR.space (language=chs, engine 2) handles complex layouts better
Configure a free OCR.space key (25000 requests/month) to enable the cloud engine and the two-engine comparison,open Settings, thenocr.spaceapply for free.