Choose a video
Pick a file from your computer. The browser reads it directly from your disk. There is no upload step.
Blur faces in any video. Your footage never leaves your device.
Drop in CCTV, bodycam or safety training footage and AI finds and blurs heads automatically. Everything runs inside this browser tab. Nothing is uploaded.
- Never uploaded
- Free, no sign-up to try
- Open-source AI
- MP4 download where supported
Video privacy blur
- 1Choose video
- 2Adjust
- 3Blur
- 4Download
Loading preview…
Preparing your video
Everything is happening on your device.
Estimating processing time…
Step 1 · Choose a video
Hi, I’m here to help. Drop in a video and I’ll blur the heads I find. It never leaves your device.
1 of 3
How it works
AI finds heads
An open-source head detector runs frame by frame on your own CPU or GPU, then blurs each head it finds.
Review and download
Compare the blurred and original video, then save your blurred copy ready to review before sharing.
Your video never leaves your browser
Sensitive footage should not have to travel to someone else’s server to be made safe. This tool does all of the work on your machine. We just made it easy.
- No upload, no server, no copy
Frames are decoded, blurred and re-encoded in this tab. inviol never receives your footage.
- Open-source models
RT-DETRv4-S head detection, published under Apache 2.0, running locally through ONNX Runtime.
- Close the tab and it’s gone
We do not store your video. The original stays on your device; keep a blurred copy by downloading it.
- Check it yourself
Open your browser’s network tab while a video processes. You’ll see the model load, and your video go nowhere.
Built for footage you need to share
- 01
Retail theft & incidents
Send CCTV to police, insurers or head office without exposing staff and bystanders.
- 02
Safety training
Turn real near-misses into toolbox talks while keeping a blameless, de-identified approach.
- 03
Investigations & reports
Attach evidence to EHS reports and board packs that meets privacy obligations.
- 04
Presentations & social
Share site footage at conferences, in decks or online without identifying anyone.
Keep data protected and identities private
Privacy is part of the wider inviol workplace safety platform, too.
Enterprise grade security
Your data is stored securely in our cloud. inviol is SOC 2 Type II certified, ISO 27001 compliant and GDPR compliant. Ensuring privacy and data access is secured to the highest standards.
Privacy first
inviol processes footage on your site in realtime, so 99% of your data never leaves site. Take a blameless approach to safety coaching with automated people and face blurring to de-identify captured footage.
Here to help
Every site is different, and our team is here to help guide you through your privacy, legal and IT Security requirements. Trusted by leading enterprises and unions from retail to aviation.
Blur every camera, automatically
inviol blurs people and faces across your existing CCTV in realtime, on site, and turns safety events into coaching moments for your teams.
FAQs
How do I blur faces in a video without uploading it?
Choose a video from your device, select a processing mode and start blurring. The AI looks for whole heads, including people facing away, and applies a strong pixel blur. Review the finished video, enter your work email and download your copy. Your video is never sent to inviol or a cloud processing service.
Is my video uploaded anywhere?
No. Your browser reads the file directly from your device and does the detection and export locally. It downloads the AI model from this website; it does not upload your video, frames, filename or detection results. We disable analytics and session recording on this tool. Your work email is sent separately when you request a download.
Can I use this for CCTV, retail incidents and safety training?
Yes. Use it to prepare short clips for a safety conversation, a training session or an incident review. Only use footage you have permission to process and share. Review the entire result: names, clothing, number plates, surroundings and other details can still identify someone.
Which accuracy should I choose?
Fast is the quickest option. Normal adds four overlapping close-ups for smaller heads. Max adds nine and is the slowest. All three use the same open-source RT-DETRv4-S head detector. These are processing presets, not different model sizes or a guarantee of detection.
Will it catch every person?
No automatic detector is perfect. Small, obscured or motion-blurred heads can be missed. Watch the complete preview before sharing. You can add fixed privacy areas that stay covered throughout the clip, or choose full-frame blur. A blurred video is not a guarantee of anonymity.
What formats work?
Use a recent Chrome, Edge or Safari with WebCodecs support. MP4, MOV and WebM input support depends on your browser’s codecs. Clips can be up to two minutes and 250 MB. Downloads are silent MP4 files where supported, with a WebM fallback, and up to 1080p for landscape footage. Processing may take much longer than the clip itself, especially on a phone.
Is it really free?
The tool is free. You can process and preview a clip before giving us any details. We ask for your work email at download so inviol can follow up about the tool and workplace video privacy. This does not subscribe you to our newsletter. The video is downloaded directly from your browser, not emailed to you.
Which AI models does it use?
Awiros’ RT-DETRv4-S person and head detector, distilled with C-RADIOv4, through ONNX Runtime in your browser. The model is published under Apache 2.0. We use head detections, not face recognition. See the model card and model and software notices. Its published benchmark is not a guarantee for your footage.
Can inviol do this across all of our cameras?
Yes. inviol blurs people and faces automatically on your existing CCTV, on site and in realtime, as part of our AI safety platform. Book a demo to see how it works.