Case study · Edge AI · A Singapore technology group, a Sri Lankan food importer and distributor, and a US AI company
VisionGuard AI at three sites in three countries
Safety and security analytics on existing IP cameras at sites in Singapore, Sri Lanka and the United States.
3countries, each on the customer's existing cameras
The problem
What the client needed
Each site already had cameras and nobody who could watch every feed: shop floors, cold rooms, loading bays and offices.
Total tickets0
No gloves0
No caps0
Customer crowd0
Employee chat0
Identified0
Violations by zone
Sweet stallSavoury section
Violations by time
08:0011:0014:0017:0021:00
No gloves · Kitchen, camera 2 · just now
The result
What runs now
Modules are switched on per camera, so each site runs only the checks it needs.
YOLO-family modelson-site GPU serverRTSP/ONVIF
What we built
The system, step by step
1
VisionGuard modules on each customer's existing IP cameras, processed on the premises
2
Two-step detection, in which a second model checks each event before anyone is alerted
3
Alerts with the clip attached, tickets with owners, and a searchable detection history
4
Detection quality measured on each customer's own cameras and agreed before go-live
Regulations and standards
PDPA (Singapore)Local data protection law
Have a problem you want measured?
Tell us what you want to count, check or protect. An engineer will reply, not a sales script.