Privacy Statement
This tool is designed to be 100% private by architecture. Here is exactly what happens — and does not happen — when you use it.
Your images never leave your device
Every step of the analysis — image decoding, OCR text detection, colour sampling, and contrast calculation — runs entirely inside your browser. No image file, pixel data, OCR result, or contrast report is sent to any server at any point.
No accounts, and privacy-friendly analytics only
There are no user accounts, no login, and no registration. This tool uses GoatCounter, an open-source, privacy-friendly analytics service, to count page views. GoatCounter sets no cookies, collects no personal data, and does not track you across other websites. It records only aggregate, anonymous information — such as the page visited, the referring site, and a rough browser and screen-size profile. Your IP address is used briefly to derive an approximate country and is never stored.
No cookies
This tool does not set any cookies.
Local storage
One value is stored in your browser's localStorage: your chosen colour theme (light, dark, muted light, or muted dark). It is used only to restore your preference on your next visit. It is never read by any server.
Network requests
Aside from the GoatCounter analytics request described above, the only network requests this tool makes by itself are a one-time download of the AI model files and the WebAssembly runtime it needs to analyse your images. These model and runtime files are served from the same origin as the tool itself (image-colour-contrast-checker.timdixon.net) and are cached by your browser after the first download.
The files downloaded on first use are:
- PaddleOCR PP-OCRv4 text detection model (
ch_PP-OCRv4_det_infer.onnx) - PaddleOCR PP-OCRv4 text recognition model (
ch_PP-OCRv4_rec_infer.onnx) - PaddleOCR character key file (
ppocr_keys_v1.txt) - ONNX Runtime Web WASM binaries
Total first-run download: approximately 44 MB. After that, your browser cache serves them instantly.
Third-party scripts
No third-party scripts, fonts or stylesheets are loaded from external origins. All JavaScript — including the GoatCounter analytics snippet (count.js) — is served from image-colour-contrast-checker.timdixon.net. The only request this tool sends by itself is the single anonymous page-view sent to GoatCounter, described above. Links to other tools send nothing unless you click them, as explained under “Links to other tools” below.
Links to other tools
Each colour combination in the report has links to two other accessibility tools: the WebAIM Contrast Checker and VADChecker (at vadchecker.com, made by Tas the Artist). Nothing is sent unless you choose to click one.
If you click, your browser opens that site in a new window. The link address contains the two colour codes for that combination, as text and background. For example, a VADChecker link ends ?txt=ffffff&bg=122222 and a WebAIM link ends ?fcolor=ffffff&bcolor=122222. The site receives those two colour codes, plus what a browser normally sends, such as your IP (Internet Protocol) address and browser details. The links are set to open with no referring address, so the site is not told which page you came from. It never receives your image, the text found in your image, or the file name.
The same links appear in downloaded PDF and Markdown reports. Anyone who clicks them later sends the colour codes in the same way.
Those sites have their own privacy practices. This tool does not control them. Please read their privacy information if you want to know how they handle data.
Open source
The full source code is publicly available under the MIT Licence. You can read it, audit it, or run your own copy. See the GitHub repository.
Contact
If you have any questions about this privacy statement, contact Tim Dixon.