Blur faces
automatically.
Upload any image β FaceShield detects every face using AI and applies a privacy blur exactly to the face shape. All processing happens locally in your browser.
Drop an image here
or click to browse your files
βΈ Console Output
Setup Instructions
01
Download face-api.js model files
Get both the
Get both the
ssd_mobilenetv1 and face_landmark_68 model weights from
the official repository:
face-api.js weights on GitHub.
02
Place models in a
Create a folder named
/models folderCreate a folder named
models in the same directory as this HTML file and copy all
downloaded weight files into it. Both ssd_mobilenetv1_model-* and
face_landmark_68_model-* files are required.
03
Serve via a local web server
Browsers block
Browsers block
file:// requests. Use: npx serve .,
python3 -m http.server, or VS Code Live Server. Then open
http://localhost:PORT/face-blur.html.
04
CDN fallback (no local setup needed)
If no local
If no local
/models folder is found, the app automatically fetches both models from
https://cdn.jsdelivr.net/npm/@vladmandic/face-api/model/. Just serve this file and it
works out of the box.
05
Tuning constants
Edit at the top of the
Edit at the top of the
<script>: BLUR_RADIUS (default 22) controls
blur strength; MIN_CONFIDENCE (default 0.4) controls detection sensitivity;
FEATHER_RADIUS (default 24) controls how soft the blur edges fade into the
original image.