Image to Text (OCR)

What does this OCR tool do?

It reads text out of images — screenshots, photos of documents, slides, receipts — and turns it into editable plain text. Recognition runs entirely on this device with an on-device OCR engine; the image is never uploaded anywhere.

How to use it

1. Drop one or more images into the upload area (JPEG, PNG, WebP, GIF, BMP, AVIF).
2. Pick a language: English, or English + Simplified Chinese.
3. Press Recognize text. Each image shows live progress and can be cancelled mid-run.
4. Edit the recognized text inline if the OCR needs a fix, then copy or download it.

Accuracy and confidence

Every image reports an average confidence percentage so you know which results to double-check. Words below a minimum confidence are dropped rather than guessed, which keeps the output clean at the cost of the occasional missing word. Clear, high-resolution screenshots give the best results; long edges above 2000 px are scaled down before recognition.

First use and privacy

The first recognition downloads about 7 MB (English) or 10 MB (English + Chinese) of OCR engine and language data from a CDN, then the browser cache serves later runs. Only the model files are downloaded — your images and the recognized text stay in this tab and are not stored anywhere.