image

Face Redactor

Detect faces locally and redact them with pixelation, blur, fills, eye bars, emoji, or custom images.

Local processing

About your filesFiles and content selected in this tool are processed in your browser and are not uploaded to a Browser Kitty backend.

Operation demo See the basic workflow in a short video.

About this tool

Face Redactor detects faces in images locally in the browser and helps anonymize people before a photo is shared. YuNet and ONNX Runtime Web are bundled into the HTML so detection and editing run on the device.

Automatic detection can be supplemented with manual masks for missed faces or any other area you want to hide. Redaction styles include pixelation, blur, solid fill, eye bars, emoji, and custom images.

Good for

Hide faces before posting photos to social media or a website
Detect many small faces in group photos and manually add any missed masks
Anonymize photos without uploading them to a remote AI service

What it can do

Key capabilities available in this tool.

Local automatic face detection
Pixelate, blur, fill, and more
Manual masks with undo / redo
JPEG / PNG / WebP export

How to use

Select an image, drag and drop it, or paste it from the clipboard.
Run face detection and review all detected masks across the image.
If small faces are missed, adjust the minimum face size or detection sensitivity.
Add manual masks where needed and choose the redaction style and strength.
Review the full image again, then export as JPEG, PNG, or WebP.

Supported

Input: file picker, drag and drop, or clipboard paste.
Detection: YuNet + ONNX Runtime Web with 12 / 24 / 48 px minimum-face presets and sensitivity controls.
Redaction: pixelate, blur, solid fill, eye bar, emoji, custom image, and manual masks.
Output: JPEG / PNG / WebP with quality/scale controls and sharing where supported.

Limitations and notes

Face detection is not perfect; missed or false detections are possible, so the full image should be checked before export.
Very small, occluded, heavily blurred, rotated, or unusually angled faces can be harder to detect.
Lowering the minimum face size can increase processing time and false positives.
Video can be loaded and previewed, but exporting a redacted video is not currently supported.

Frequently asked questions

Are photos sent to a server?

No. The detection model and runtime are bundled in the HTML, and image loading, detection, and editing run in the browser.

What if automatic detection misses a face?

Add a manual mask, or lower the minimum face size to 12 px and run detection again for small faces.

Can I redact areas other than faces?

Yes. Manual masks can cover any area, such as a license plate or name badge.

Which output formats are supported?

JPEG, PNG, and WebP are supported, with output quality and scale controls.

Offline version and source code

The root-level face-redactor.html on GitHub is an expanded single HTML file with ONNX Runtime Web and the YuNet model bundled inside. You can download and open it directly in a browser, with face detection and image editing performed locally.

View source on GitHub