noRecognition Tester Open the tester
You are the test

Find out if the cameras
can still see you.

Point a camera at yourself, wearing the pattern, and watch a real surveillance model try to find you. No lab, no middleman, no trust required. The detector runs on your own device, and you read the result with your own eyes.

Private by construction not by promise

Runs on your device

The detection model is downloaded once and runs inside this browser. Your camera feed is never uploaded, never proxied, never stored.

Nothing is tracked

No analytics, no advertising, no cross-site cookies. Your camera and your results stay on your device: we only see how a test went if you choose to send us that session.

Yours to erase

Recordings and your face data live in this browser only, and one tap deletes them for good. Sign out and the models stay cached for next time.

The one exception, and only if you choose it: a session records to your own device as you test, so you can review it or, if you want to, send it in. Nothing is transmitted on its own. On the stop screen you pick one of three: send nothing, send the data only (detections, confidences, heatmaps), or send the data and the video. You see exactly what would go before anything leaves your device, and you can have it deleted later.

What this tool actually does

A modern camera does not "see" a person. It runs the picture through a detection model that outputs a confidence score: how sure it is that a person, a face, or a specific identity is present. This tester runs those same models, in your browser, so you can watch the score for yourself with and without the pattern.
The honest bar is simple. A pattern only counts if the model can find you in plain clothing and then cannot find you wearing the print, at the same distance and the same light. The tool enforces that: it refuses to compare two takes that do not match, because distance and brightness move the score on their own.

Exactly what we have, and what we don't

No vague promises. Here is the complete list of what reaches our server, and why each thing is there.
Your email address. You sign in with it, so of course we have it. It is used to send your sign-in link, to enforce your account's fair-use limit, and to reply to you if you send us a diagnostic. It is never sold, never shared, and never used to track you anywhere else.
A tally of which models your account has downloaded. The detection models are licensed, so we keep the model name and the date. That is all it is for: keeping the free limit fair and shutting off abuse. It records nothing about your results.
Only the sessions you choose to send us. Your session records to your own device automatically, and stays there. When you decide to send one, you choose whether it is the data only (detections, confidences, heatmaps, and the session details) or the data and the short video, and it is tied to your account on purpose: we need it to see the problem and to write back to you, and it is what credits your contribution. You preview exactly what will be sent, nothing leaves your device until you approve it, and you can have it deleted afterward.
What never reaches us at all: your live camera while you test (it is processed in your browser and discarded frame by frame), your results unless you choose to send a session, and your face (if you enroll one for the recognition test it stays in this browser as a list of numbers, never uploaded, never part of a diagnostic). No analytics, no advertising, no cookie that follows you around the web.