Image

Image to Text (OCR) (β)

Image to Text is an OCR (optical character recognition) tool that detects the text inside an image and pulls it out as editable text. Upload a photo, screenshot, or scanned document, and it recognizes the characters and turns them into text you can edit and copy. No sign-up required, and it's free to use online.

Drag and drop an image here

Supports JPG, PNG, WebP, and BMP. One image at a time (up to 25 MB).

Images are processed entirely in your browser and never sent to a server. Runs entirely in your browser.

Only the data needed for text recognition is loaded over the internet on first use, then stored in your browser and reused. Your images are never sent anywhere.

You can also paste from the clipboard with Cmd / Ctrl + V.

Image Compressor

Reduce JPG / PNG file size in your browser (quality slider for JPG, lossless optimization for PNG).

Image Cropper

Crop JPG / PNG / GIF images in your browser with aspect ratio presets or precise numeric coordinates.

WebP Converter

Convert JPG / PNG / SVG to lightweight WebP in your browser (batch up to 10 files, ZIP download supported).

Image Resizer

Resize JPG / PNG / GIF / WebP images in your browser by pixel, by percent, or with aspect ratio locked.

Slack Custom Emoji Generator

Create Slack custom emoji (128×128 / 128 KB) from text or images entirely in your browser (fonts, colors, alignment, outline, GIF animation supported).

About Image to Text (OCR) (β)

This tool is currently in beta. It reaches good accuracy on clean material such as document photos, scans, and screenshots, where the text is crisp on a plain background, but accuracy may be insufficient on photos with busy backgrounds, decorative fonts, or skewed and blurry images.

It supports Japanese and English plus many other languages, including Simplified Chinese, Traditional Chinese, Korean, Indonesian, French, German, Spanish, and Portuguese, and you can select several languages at once for mixed-language images. The result is editable on the spot, so you can fix any misreads before copying it or downloading it as a .txt file. For images with unusual layouts such as signs, tables, and receipts, you can switch the layout analysis mode.

Every image you upload is processed entirely in your browser and never sent to an external server (only the data needed for text recognition is loaded over the internet on first use, then stored in your browser and reused). That makes it safe to extract text from confidential documents or images containing personal information. Runs entirely in your browser.

How to use

Drag and drop an image onto the drop area, or click to choose a file (pasting from the clipboard also works).
Pick the languages you want to recognize on the right (Japanese and English are selected by default; you can choose multiple).
If needed, switch the layout analysis mode or toggle "Preserve line breaks."
Press "Extract text" to start recognition; the extracted text appears on the right.
Edit the result if needed, then take it out with "Copy" or "Download .txt."

Use cases

Business users who want to read and digitize text from photos of paper documents or whiteboards.
Pulling text out of a screenshot when you can't select or copy it, so you can quote or transcribe it.
Converting business cards, receipts, and slips to text from an image instead of typing them by hand.
Running multilingual OCR on foreign-language signs, menus, or documents before sending the text on to a translator.
Editors and writers who photograph pages of old books or brochures and want to read and reuse the text.

Notes

This tool is currently in beta. It works best on images with a clean background and crisp text, such as document photos, scans, and screenshots. Accuracy may be insufficient on photos with busy backgrounds, decorative fonts, or distorted text.
One image at a time, up to 25 MB per file (JPG, PNG, WebP, and BMP supported).
Accuracy drops for handwriting and for blurry, skewed, or low-resolution images. Sharp, level images give the best results.
The fewer languages you select, the more stable the accuracy and speed. Choosing too many at once can increase misreads.
The data needed for text recognition is loaded over the internet on first use only, then stored in your browser and reused, so the first run takes a little longer to load.
OCR isn't perfect. Review the extracted text and correct any misreads before using it.

FAQ

Are my images uploaded to a server?

No. All text recognition happens entirely in your browser. Only the data needed for text recognition is loaded over the internet on first use, then stored in your browser and reused, but the image itself is never sent anywhere. That makes it safe to use with confidential documents or images that contain personal information.

Can it read handwriting?

Accuracy for handwriting is much lower than for printed type. It can sometimes read it, but cursive and joined-up writing are difficult. For best results, use neatly written text or printed type.

How can I improve recognition accuracy?

Use a sharp, in-focus image and capture or crop it so the text sits level. Limiting the languages to only those that actually appear in the image also reduces misreads. For unusual layouts such as signs or tables, try switching the layout analysis mode.

Which languages and image formats are supported?

It supports Japanese, English, Simplified Chinese, Traditional Chinese, Korean, Indonesian, French, German, Spanish, and Portuguese, and you can select several at once. Supported image formats are JPG, PNG, WebP, and BMP.

Can I use the extracted text as-is?

You can edit the result right on the screen. Because OCR can misread characters, review the text and correct it as needed before copying or downloading it as .txt.