Image to Text (OCR)
Reads the text in a photo, scan or screenshot, in your browser, in over 30 languages.
Runs in your browser. The file or text you give it is not uploaded. Free, no account.
How it works
You have a screenshot, a scan or a photo with text you need to type out by hand. Drop the image in. Tesseract runs in your browser, reads the text in the language you pick, and gives you the words to edit and copy, with a confidence figure. The image is not uploaded.
- 1
Drop in an image. A text-recognition engine (Tesseract) runs inside your browser and reads it; the image is not uploaded.
- 2
Choose the language of the text. English is ready at once; other languages download their data the first time (a few megabytes).
- 3
You get the text to edit and copy, with a confidence figure.
What it does not do
Sharp, straight, high-contrast images read best. Handwriting and stylised fonts read poorly.
Layout such as columns and tables is flattened into lines of text.
When a source is busy or a site does not answer, this tool says so. It never fills the gap with a guess.
Questions
Why is the OCR result so poor?
Text recognition works well on sharp, straight, high-contrast images. Blur, a tilted page, shadow or low contrast all cost accuracy. Handwriting and stylised fonts read poorly, so do not expect clean output from them. Straighten the shot, raise the contrast, crop to the text, and run it again.
Does my image get uploaded to a server?
No. The recognition engine runs inside your browser, so the image stays on your machine and is not uploaded. When you pick a language other than English, its data (a few megabytes) downloads the first time you use it. English is ready at once.
Can it read the text in columns or a table?
It returns lines of text, not the original layout. Columns and tables come out flattened, so a table may read across rows and columns in a way you have to sort out by hand. For a small amount of text that is fine. For a long table, expect to tidy the result in your own editor.