Privacy Policy

Welcome to Umi OCR (umiocr.com). Your privacy is important to us. This policy explains what information we collect, how we use it, and how we protect your data.

1. How OCR Processing Works

All OCR processing runs entirely on your device. When you upload an image or PDF, it is processed locally using ONNX Runtime Web — no image data is ever transmitted to our servers or any third party. Your files never leave your device during OCR.

  • No uploads: Images and PDFs you process are handled entirely on-device and are never sent to any server.
  • No storage: We do not store, log, or retain any image, PDF, or extracted text content.
  • No account required: You do not need to register, log in, or provide any personal information to use the OCR tool.

2. Information We Collect

Umiocr collects minimal information to operate the website:

  • Feedback form: If you voluntarily submit feedback, we collect the name, email address, and message you provide in order to review and respond to your request.
  • Abuse prevention: Feedback submissions may store a hashed version of your IP address to prevent spam and repeated submissions.
  • Analytics: We may use standard web analytics (such as page views and referral sources) to understand site traffic. This data is aggregated and not linked to personally identifiable information.

3. Cookies

Umiocr may use basic cookies to support site functionality and analytics. Our third-party advertising provider may also use cookies or similar technologies to deliver and measure ads. Cookies do not contain any image data or OCR results. You can disable cookies in your browser settings; doing so will not affect the OCR tool.

4. Advertising

Our pages display advertising supplied by a third-party ad network. The ad provider may receive technical information such as your IP address, browser details, and page URL, and may use cookies or similar technology when its ad loads. Ads appear in a separate browser frame; your uploaded images, PDFs, and recognized text are not sent to the ad provider by the OCR tool. You can close the ad banner at any time.

5. Open Source Model Disclosure

Umi OCR uses the PP-OCRv6 Small and Tiny model weights released by PaddleOCR under its open source license. These model weights are loaded into your browser via ONNX Runtime Web and run entirely on your device. Umiocr is an independent project and is not affiliated with or endorsed by PaddleOCR or PaddlePaddle.

When browser storage is available, Umi OCR saves the downloaded public model files in your browser so they can be reused after a refresh or on another OCR tool page. This cache contains models only, not your images, PDFs, or recognized text. You can remove it by clearing this site's stored data in your browser settings; the models will then be downloaded again when needed.

Models are downloaded from the official distribution first. If that download fails, the tool tries our Cloudflare R2 mirror of the same version and verifies the file before use. These downloads do not include your documents or recognized text.

6. Consent

By using this website, you consent to this Privacy Policy and agree to its terms.

7. Contact

If you have questions about this Privacy Policy, please reach out via our public repository or domain administrator.

← Back to Homepage