Image to Text
Read printed text from a PNG, JPEG, or WebP image on your device. Edit the recognized text and download a text file.
Clicking Read text downloads the selected language model from tessdata.projectnaptha.com if it is not cached. The first run needs an internet connection and may download tens of megabytes. Your image and text stay on this device.
About Image to Text
Read printed text from a PNG, JPEG, or WebP image on your device. Edit the recognized text and download a text file.
Frequently Asked Questions
No. Your browser runs Tesseract OCR locally. On first use of a language it downloads a model from tessdata.projectnaptha.com, which receives a normal download request and your IP address, but no image or recognized text. The engine loads from this website. Models may be cached in your browser for later use.
Clear, straight photos or screenshots of printed text work best. Select the language used in the image. Handwriting, curved pages, complex layouts and low contrast can produce mistakes. Review names, dates and numbers yourself; the confidence score is an estimate, not a guarantee.
One PNG, JPEG, or WebP at a time, up to 12 MB and 6 megapixels. The first frame of animated images is used. Recognition stops after 3 minutes and can be cancelled. Large models may use substantial memory, so close unused tabs on small devices. PDFs should first be exported to images.