100% on-device processing
Anonymize faces and bodies without uploading anything
Detect and mask people in images using on-device machine learning. Video support is in Beta - your media never leaves your browser.
Built for privacy
Everything runs locally. Your media never leaves your device.
On-device ML
MediaPipe models run in your browser via WebAssembly and WebGPU. No cloud APIs.
Auto & manual tracking
Detect faces and bodies automatically, then fine-tune masks or track them across video frames.
Irreversible masking
Solid fill permanently destroys pixel data. We explain the security trade-offs of every method.
Not all blur is equal
Modern machine learning tools can reverse common techniques. Choose based on your security needs.
Solid fill
Replaces pixels with a solid color. Original data is permanently destroyed.
Pixelate
Reduces resolution to large colored blocks. Information is aggregated and lost.
Use block size of 20px+ for security. Smaller blocks may be partially reversible with machine learning tools.
Gaussian blur
Applies mathematical smoothing. Research shows machine learning can reconstruct blurred faces.
Cosmetic blur only. NOT SECURE. Machine learning can reverse this effect.
Your data stays yours
Never trust a server with sensitive media. Detection models run entirely in your browser.
- No file uploads. Media stays in browser memory
- ML inference runs via WebAssembly and WebGPU
- Exports are generated locally and downloaded directly
- Works offline after models are cached