Automatically blur faces and plates in videos
Detect and blur faces and licence plates automatically, entirely in your browser. No uploads, no account — your footage never leaves your device.
Free with a small watermark · £10 one-off to remove it
Why choose AnonymizeVideo.app?
Videos never uploaded
Every frame is processed on your own device. Your footage is never sent to a server or stored in the cloud.
Fast GPU processing
Hardware acceleration via WebGPU is used automatically wherever your browser supports it, with a WASM fallback.
Automatic face and plate detection
Detection models find and track faces and licence plates for you — no manual, frame-by-frame editing required.
No account required
Open the page and start blurring. No sign-up, no install, no subscription to get going.
How it works
-
Upload
Drop in an MP4 or MOV. It loads instantly and stays on your device.
-
Detection finds faces and plates
The detector scans your video and tracks faces and licence plates across frames.
-
Review detections
Check each face, change the blur style, or remove any false positives.
-
Export MP4
Render a redacted video and download it — encoded locally, never uploaded.
Who uses it
Dashcam videos
Blur bystanders and plates before sharing footage of an incident, a near-miss, or a drive.
YouTube creators
Anonymize members of the public so you can publish vlogs without consent forms.
Journalists
Protect sources and bystanders in sensitive footage without using a cloud service.
Schools & education
Share classroom and event recordings while protecting students’ privacy.
Your video never leaves your computer.
Unlike cloud video editors, AnonymizeVideo.app processes every frame locally in your browser using WebCodecs and WebAssembly. Sensitive footage is never uploaded, queued on a server, or stored anywhere but your own device.
- Runs entirely in your browser
- Powered by ONNX Runtime
- Supports hardware acceleration
- Open standards: WebCodecs + WebAssembly
How automatic face and licence plate blurring works
Automatic video redaction uses detection models to find faces and vehicle plates in each frame, then applies blur, pixelate, or a solid block over each one. AnonymizeVideo.app runs these models directly in your browser. For people, choose a face model. For vehicles, choose the License plates model, which covers licence plates and number plates alike. As your video moves through the pipeline, detections are tracked so redactions stay attached even when subjects move or are briefly obscured. When you export, the app re-renders every frame and encodes to MP4 without a single upload.
When should you anonymize video?
Any time footage contains identifiable people or vehicles, it is worth blurring before you publish or share it. Common cases include dashcam clips of an incident, street interviews and vlogs uploaded to YouTube or social media, journalistic material that must protect sources and bystanders, research recordings, and classroom or event videos shared by schools.
Redacting faces and plates also helps you stay aligned with privacy expectations and data-protection rules such as the GDPR, which treat a recognisable face or registration plate as potentially identifying data. Because the detection is automatic, anonymising a clip takes minutes rather than hours of manual editing.
Why browser processing is more private
Most online tools that claim to blur faces upload your video to a server, process it in the cloud, and send the result back to you. That means your footage — which may show identifiable people, locations and vehicles, along with all the context around them — leaves your control and sits on someone else’s infrastructure.
AnonymizeVideo.app is different: it never uploads anything. Modern browser technologies make this possible. WebCodecs decodes and re-encodes the video using your device’s own media hardware, and WebAssembly runs the detection model at near-native speed. Once you have run one scan the app, its runtime and the model are all cached, so you can disconnect from the internet entirely and still process video — proof that nothing is being sent anywhere.
Supported video formats and limits
MP4 and MOV files work best as input. To blur plates, select the License plates model from the Detection model dropdown before scanning. When you export, you can choose H.264 / AVC for the broadest playback compatibility, VP9 for smaller files, or AV1 for the smallest files at the cost of slower encoding. You can preserve the original audio track, scale the output resolution down to reduce file size, and control the target bitrate.
Because the finished video is assembled in memory before it downloads, very long or very high-resolution files are limited by how much RAM your device has rather than by any server quota. Around 15 minutes at 1080p is comfortable on typical hardware; beyond that, scaling the output resolution down makes a long job far more likely to finish.
Frequently asked questions
Is my video ever uploaded to a server?
No. AnonymizeVideo.app runs entirely in your browser. Your file is decoded, analysed, redacted, and re-encoded on your own device — it never leaves your computer and is never sent anywhere.
How does face and license plate detection work?
Detection models run locally via WebAssembly (and WebGPU where available). You can detect faces or switch to the license plate model for licence plates and number plates, then apply blur, pixelate, or a solid block to each detection.
How do I blur licence plates, license plates, or number plates?
In the Detection model dropdown, select License plates, then run the scan. The app will detect vehicle plates so you can blur, pixelate, or block them before export.
What video formats are supported?
MP4 and MOV files work best. You can export with H.264 for maximum compatibility, or VP9 / AV1 for smaller files. Original audio can optionally be preserved.
How long a video can I process?
Around 15 minutes at 1080p is comfortable on typical hardware. There is no server quota — the limit is your own device’s memory, because the finished file is assembled in RAM before it downloads. For longer or higher-resolution footage, scaling the output resolution down in the export options makes the job far more likely to finish.
Which browsers does it work in?
Any browser with WebCodecs support: Chrome and Edge, Safari 17 and later, and recent versions of Firefox. The page checks when it loads and tells you if something is missing, so you find out before choosing a file. A desktop or laptop is much more dependable than a phone for anything longer than a short clip.
What if it misses a face or blurs too much?
After the scan you can review every detected face or plate individually, change or remove its redaction style, and fine-tune the confidence threshold or region size before exporting. Lowering the threshold and re-running catches harder detections.
Is it free, and do I need an account?
Yes, it is free to use without an account, however a small watermark is applied to the video. To remove the watermark you can pay £10 to upgrade, which requires registering an account.
Does it work offline?
After your first scan the app, its detection runtime and the model are all cached, so it runs with no network connection at all. The scan is what matters rather than the visit: the runtime and the model are downloaded the first time something needs them, so loading the page alone does not fetch them. Once you have run one scan you can work offline to be certain your video is not leaving your device.