Malaysia's largest shopping centre put self-learning AI across the CCTV network it already had. Crime fell 58.8% over five months. On-ground security staff came down 13% in the first quarter. The team explains how, in three minutes.
A working session on your own camera network, with a security engineer who has deployed in malls this size.
Filmed on site with AE Security Systems and the 1 Utama security team.
Covering 1 Utama takes more than 1,000 cameras. In a conventional control room nobody watches all of them, so most incidents only surface once somebody reports them. Three things changed when self-learning AI went live.
Operators used to replay footage after an incident to work out what had happened. Now unusual activity is flagged as it occurs and the nearest guard is dispatched straight away.
icetana watches every camera and works out what normal looks like on each one. It sharpens the longer it runs, so there is no rule library to maintain and no tuning per camera.
AE Security cut on-ground staff by 13% in the first quarter, with a further 7% planned. Operators now spend their time resolving incidents rather than scanning feeds.
What impressed us the most about the product is that it uses a self-configuring and self-learning AI to detect unusual events. Essentially, the longer you use the system, the better it will detect any anomalies in the environment.
With AI, the control room detects events or emergencies immediately when they happen. Then they will dispatch the nearest security guard to attend to the issue. This has helped us improve response times tremendously.
If you run a shopping centre, a mixed-use precinct or a portfolio of buildings across Asia Pacific, the problem is the same. There are more cameras than anyone can watch, and the control room is measured on how fast it responds.
icetana reads your existing camera streams and sits alongside your video management system. Genetec Security Center and Milestone XProtect are supported integrations.
Over its first days of running, the system builds its own picture of normal movement on every camera. Nobody draws zones or writes rules to get it started.
Unusual activity surfaces on the wall as it happens: loitering, aggressive movement, falls and medical emergencies, plus intrusion across any perimeter you choose to draw.
Tell us about the site and we will set up a 30 minute session with a security engineer. You will see how the AI behaves on a network the size of yours, what it surfaces in the first week and what the licence looks like.
