🙈Face Blur (Auto)

A lifestyle photo shows a hoodie, a model and two passers-by. A review photo shows a customer's kitchen and a customer's children. A warehouse video still shows your staff. Publishing any of those without consent is a legal risk, and blurring each face by hand across a catalogue is tedious. The Face Blur (Auto) tool detects and blurs every face automatically, so you can anonymise photo sets and blur faces in bulk in the time it takes to drag a folder. It is an auto blur faces tool with nothing to draw and nothing to upload.

How it works

1. Drop the images, up to 200 at a time.
2. On first use the tool loads the MediaPipe BlazeFace detector, about 12 MB, once; your browser caches it for later visits.
3. Every face is detected, including small faces in the background and faces at an angle, and a blur, pixelation or solid patch is applied to each box.
4. Adjust padding to extend the patch beyond the detection box, so hairlines, ears and the edge of the jaw are covered too, and set the blur strength.
5. Review the previews, then download individual files or the ZIP.

Detection runs on a resized copy for speed, but the face blur is applied to the full-resolution pixels so the result stays print quality. A count of detected faces appears on each card, which is a quick way to spot an image that needs a manual pass.

Why it matters

Faces are personal data. In the EU and UK under GDPR, and in a growing list of other jurisdictions, publishing a recognisable person for commercial purposes without a lawful basis is a violation, and penalties are calculated per incident. Marketplaces add their own layer:

• Amazon lifestyle images and A+ content routinely use models; a face without a model release is a takedown risk.
• Etsy and Shopify stores that showcase customer photos need consent or anonymisation for each one.
• TikTok Shop creators frequently film in public; bystanders in cover images should be blurred before the still is reused as a product image.

Beyond compliance, blurring bystanders keeps attention on the product, which is what the image was for.

Tips

• Set padding to 25 to 40%; the detector box is tight around the eyes, nose and mouth, and an unblurred forehead or chin is still identifying.
• Choose pixelate with large blocks for real anonymisation; a light blur can be partially reversed.
• Check crowd scenes manually. The detector is excellent on faces above about 20 px, but a distant profile can be missed; add a region with the Blur / Pixelate Region tool for anything it skips.
• Do not run this on passport or ID photos by accident; the Passport Photo Maker is the tool for those, and it uses the same detector to centre the face rather than hide it.
• Strip the location and timestamp with the EXIF / Metadata Remover after blurring; a blurred face with GPS coordinates is only half anonymised.
• Where the person is the point of the photo, consider removing the background with the Background Remover instead, which also removes bystanders.

Privacy

The face detector runs on-device. Your images never leave your browser, no faces are sent anywhere for analysis and no biometric data is stored.