Browser Kitty
日本語
video

Video Face Redactor

Detect faces in video locally, redact them, and export an H.264 MP4.

View on GitHub
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

Video Face Redactor detects faces in video locally, covers them with mosaic, blur, solid blocks, or emoji, and exports an MP4. YuNet and ONNX Runtime Web handle face detection while review, correction, masking, and encoding all stay inside the browser; the selected video is not uploaded.

The workflow includes frame-by-frame review, re-detection for a specific scene, manual masks, and adjustable padding so you can correct misses instead of relying solely on automation. H.264 input is supported, and H.265/HEVC can also be accepted when the browser, OS, and device provide compatible WebCodecs decoding.

Good for

  • Hide bystanders or family members before posting a video to social media
  • Anonymize faces in event, school, or community footage before sharing
  • Redact faces without uploading the source to an online video editor

What it can do

Key capabilities available in this tool.

How to use

  1. Open the tool and choose an MP4 or MOV video.
  2. After automatic face detection, play the video or step one frame backward/forward to review mask positions.
  3. Re-detect faces in scenes where masks drift and add manual masks for missed faces.
  4. Choose mosaic, blur, solid fill, or emoji and adjust padding around faces.
  5. Confirm filename, resolution, quality, and audio options, export the H.264 MP4, and review the entire result before sharing.

Supported

  • Input video: H.264/AVC; H.265/HEVC when WebCodecs decoding is available on the browser/OS/device
  • Input containers: primarily MP4 / MOV
  • Output: H.264 MP4 at original resolution, 1080p, or 720p
  • Compatible AAC audio can be copied without re-encoding

Limitations and notes

  • Face detection and tracking are not perfect. Review the complete exported video before sharing and add manual masks when necessary.
  • HEVC input and H.264 export depend on WebCodecs support in the browser, OS, and device.
  • AV1 and VP9 input are currently outside the supported scope.
  • Unsupported audio may be omitted after a warning, and long or high-resolution videos can require substantial processing time and device resources.

Frequently asked questions

Will every face be hidden automatically?

No guarantee is possible. Review the detections, re-detect problematic scenes, add manual masks for misses, and watch the full exported video before sharing.

Can I use H.265/HEVC phone videos?

Yes when the current browser, OS, and device expose compatible HEVC decoding through WebCodecs. The app checks support when you select the file.

Is the video uploaded to a server?

No. Detection, review, correction, masking, and MP4 export are performed locally in the browser.

Offline version and source code

The standalone HTML bundles the face model, ONNX Runtime Web, and required runtime code so processing can remain local in a compatible browser.

View source on GitHub

Use together

Useful next steps and companion tools, organized around the workflow.

Compress next Video Compressor Compress the redacted video to a more shareable file size. Inspect output Media Inspector Inspect codec, resolution, audio, and other output details. For photos Face Redactor Use the lighter image tool when the source is a photo rather than a video.