Free · no signup · browser-local
Image to Text OCR: English and Simplified Chinese
Extract printed English or Simplified Chinese text from PNG, JPEG or WebP locally. Adjust rotation, scale and threshold, review the image and edit text before download.
Browser-local OCR · image and recognized text stay in memory · engine files load from this site
Image and settings
Choose or drop one PNG, JPEG or WebP. Up to 15 MiB / 16 million pixels.
Preprocessing is optional and can reduce accuracy. 2× output must also fit the 16-million-pixel limit. Rotation is applied to the original pixels; EXIF orientation is not used.
Recognized text
Choose an image to begin. The engine loads only when you press Recognize text.
Best suited to clean, printed English and Simplified Chinese. Handwriting, curved pages, tiny text and complex layouts can produce errors. No cloud model is used.
Getting started
Choose PNG, JPEG or WebP, choose language and layout, inspect the preview and start recognition. Review and edit the text before downloading TXT or JSON.
Before you use the result
OCR can misread or omit characters. This is local recognition, not translation or handwriting expertise. One image up to 15 MiB and 16 million pixels; program/model files load from this site on first use.
Try an example
Choose a clear printed English or Simplified Chinese sample image, run OCR, and compare every recognized line with the image.
How it works
Tesseract.js 7.0.0 in a browser Worker with self-hosted WASM and pinned English/Chinese models. Optional scaling, rotation and grayscale/threshold processing run locally; confidence is an engine estimate.
Frequently asked questions
Recognize English or Simplified Chinese text from a local image, with a preview, preprocessing, progress and editable output.
How do I use Image to Text OCR?
Choose PNG, JPEG or WebP, choose language and layout, inspect the preview and start recognition. Review and edit the text before downloading TXT or JSON.
Does this tool upload my data?
Inputs and files are processed in your browser. Page-view analytics do not contain your tool input.
What should I check before using the result?
OCR can misread or omit characters. This is local recognition, not translation or handwriting expertise. One image up to 15 MiB and 16 million pixels; program/model files load from this site on first use.