On-site video verification stops tailgating without compromising member privacy - Gatech

On-site video verification stops tailgating without compromising member privacy

Why entrance video should be checked on site, not somewhere else

A scan tells you a credential was accepted. It does not tell you how many bodies walked through. Checking entrance footage on the premises closes that gap in seconds, and the material never has to leave the building to do its job.

Counting people is a separate job from checking credentials

Picture a Tuesday at 18:40 in an unmanned club. One member taps in. Two people walk in. The access event looks perfectly normal, because it is normal — the credential was valid. The problem is the second body.

A verification layer compares what happened at the entrance against the entries the system accepted. When the numbers do not match, that moment gets flagged. A human looks at it later and decides. The machine does not name anyone, and it should not pretend to. It only points at a mismatch and shows the few seconds that matter.

This is also why the physical barrier still carries the weight. Software decides who is allowed in. A turnstile decides how many bodies actually fit through one accepted entry. Without that, you are reviewing a steady stream of flagged events instead of rare ones, and review time becomes a staffing cost.

Local processing keeps it fast and keeps it quiet

Reviewing on site has two business effects. Decisions happen now, not after a delay, so a pattern of shared access is caught in days rather than at the end of a quarter. And the footage stays where the member expects it to stay — on the premises, under your retention rules, visible only to the people you authorise.

Nordic members accept reasonable checks. They are less relaxed about their image travelling somewhere they cannot see. Keeping the review local removes that objection before anyone raises it.

It will not be perfect. Crowded moments, coats, prams — some events will be ambiguous. The honest goal is fewer unverified entries, not zero.

Frequently asked questions

Does this recognise who the person is?

No. The layer compares how many people passed against how many entries were accepted. Identifying someone is a human decision made during review, not something the equipment claims to do.

How much staff time does the review actually take?

Only flagged moments are looked at, not full days of footage. In a club with a proper physical barrier, that is usually a handful of short clips per week.

We already have a barrier. Do we still need verification?

A barrier stops most multi-person entries, but not all of them — people hold, push and follow. Verification shows you the ones that got through, so you know whether the problem is three a month or thirty.

What happens to the footage?

It stays on site and is deleted according to the retention period you set. Because nothing is sent elsewhere for checking, the answer you give a member is short and verifiable.