How JustScanText Works
Your images are processed entirely in your browser. Here is exactly what happens under the hood.
Browser-Side Processing
JustScanText runs entirely in your web browser. When you load an image, every step — image decoding, OCR analysis, and text output — happens on your device. No file is ever sent to any server, and no external service processes your data.
Technical Implementation
The tool is built on these technologies and open-source libraries:
Tesseract.js
An open-source OCR engine compiled to WebAssembly. It runs Tesseract — the same OCR engine historically used by Google — entirely inside your browser. Language model files (~1–10 MB each) are downloaded on demand and cached locally.
WebAssembly (WASM)
The technology that allows compiled C/C++ code to run at near-native speed inside your browser. It is the foundation that makes browser-based OCR practical — no plugins or server round-trips required.
SharedArrayBuffer + Web Workers
JustScanText uses multi-threaded processing via SharedArrayBuffer and Web Workers for faster OCR. This requires Cross-Origin Isolation headers (COOP/COEP), which is why the OCR engine runs on a separate subdomain (tool.justscantext.com).
File API & Blob URLs
Images you drop are read directly into browser memory via the File API. No network request is made at any point during OCR processing.
What This Means for Your Privacy
- Your images are never uploaded to any server
- No third party receives or processes your image data
- We have no access to the content of your files or the text extracted from them
- No file data is logged or stored anywhere
- Files exist only in your browser's memory during processing
How to Verify
You do not have to take our word for it. Here is how to confirm:
- 1 Open your browser's developer tools (press F12, or right-click and choose Inspect)
- 2 Go to the Network tab
- 3 Use the tool to load and scan an image
- 4 Observe: no outbound requests contain your image data
Alternatively, disconnect from the internet after the page loads. The OCR tool continues to work — because it requires no server.
Trade-offs and Limitations
Because processing happens on your device, performance depends on your hardware. Very large images or batch operations may take longer on slower or low-memory devices than server-based tools. Language model files need to be downloaded the first time you use a non-English language. These are the trade-offs for keeping your files completely private.