AI OCR: Extract Text From Any Image
Pull the text out of any image, screenshot, receipt, sign, slide or scan, verbatim, with AI OCR running entirely in your browser. Drop an image in and the tool reads it, then draws a clickable box over every line of text it found. Click any box (or its row in the side list) to copy just that line, hover a row to highlight its box on the image, or grab everything at once with Copy all text. You also get a live character and line count and a one-click Download .txt. The OCR model (Florence-2) loads once on first use, then is cached, and because everything runs locally your image is never uploaded.
How to extract text from an image
- Drop an image or screenshot in, or click to browse.
- Wait a few seconds while the AI reads the text and maps each line to a box.
- Click a box to copy that line, or use Copy all text or Download .txt.
Examples
Screenshot to text
A screenshot of an error dialog with a long message and a code
Each line boxed on the image; click the code to copy just it, or copy the whole message at once.
Receipt to text
A photo of a printed receipt
Every line item read out as text, ready to copy or download as a .txt file.
Frequently asked questions
Is my image uploaded anywhere?
No. The OCR model runs entirely in your browser, using your GPU via WebGPU when available and WebAssembly otherwise. Your image is read on your device and never uploaded. Only the model itself downloads, once, then is cached for instant reuse.
Which model powers the text extraction?
Florence-2, a compact vision model from Microsoft (MIT licensed). It reads the text and also returns a bounding box per line, which is what lets the tool overlay clickable regions and copy a single line. It runs locally in your browser, so nothing leaves your device.
What kinds of images work best?
Clear, high-contrast text reads best: screenshots, slides, signs, labels, receipts and document scans. Photos work too, but blurry, low-light, heavily stylized or handwritten text is harder. If a line is missed, try a sharper or higher-contrast image.
Can I copy just one line instead of everything?
Yes. Every detected line is drawn as a box on the image and listed in reading order beside it. Click a box or its row to copy just that line (you get a brief copied confirmation). Hovering a row highlights its box so you can see exactly where it sits. Copy all text grabs the full result, and Download .txt saves it to a file.
Why does the first run take a moment?
The Florence-2 model downloads the first time you use the tool, then is cached for instant reuse on later images. After that, OCR is fast and fully offline. Larger images also take a little longer because more pixels are analyzed.
Related tools
Image to Text (OCR)
Extract text from an image with OCR, right in your browser. Supports many languages, copy or download the result. Your image is never uploaded.
AI Photo Tagger
Drop a photo and get an AI caption plus clean keyword tags, in your browser. Copy as a comma list or hashtags for stock sites, SEO, and assets.
Object Detector
Detect objects in any photo with YOLO11, right in your browser. Draw labeled boxes, count what it finds, and download the result. Nothing is uploaded.
Add Border to Image
Add a colored border or frame around an image, with adjustable width and color. Live preview, runs in your browser, nothing uploaded.
AI Alt Text Generator
Generate accessible alt text for any image with AI, right in your browser. Copy clean alt text or a ready HTML img tag. Nothing is uploaded.
AI Background Remover
Remove the background from a photo automatically with AI, right in your browser. Get a transparent PNG or a solid color. Nothing is uploaded.