University of the Western Cape Tests Facial Recognition Infrastructure
The University of the Western Cape recently tested facial recognition technology integrated directly into its campus CCTV network. The five-day pilot used software from Swiss-South African firm SAFEZA and generative AI inference from Johannesburg-based Gigabrain to process student access.
Swiss-South African Technology Integration Details
The technical session and live deployment took approximately two hours to complete. Implementation relied on a secure Tailscale virtual private network built on the WireGuard protocol. This setup connected multiple devices into a single encrypted network so administrators could manage remote access. The deployment processed video streams from the university’s three main camera networks as thousands of students accessed the campus daily. AI inference ran locally in South Africa through Gigabrain rather than via an overseas cloud environment, preventing perceptible delays in identifying individuals on campus.
System uses facial vectors for watchlists and access
University officials discussed potential applications including automated watchlists to flag unauthorized individuals or former employees. The system alerts security personnel so staff can escort flagged persons from the grounds. Organisers also explored using facial recognition to replace physical student cards for entrance and exit events. SAFEZA stated that the pilot involved informed participants who were aware of the facial recognition system’s deployment. The system converts facial characteristics into a mathematical representation called a facial vector rather than storing photographs.
Risk director praises accurate identification results
Bently Ebrahim Jeftha, director of risk and compliance at the University of the Western Cape, described the setup process as effortless. He noted that the facial recognition functionality performed well during the proof of concept. Jeftha highlighted accurate identification results and a very low incidence of false readings during the trial.
University considers permanent rollout with expanded detection features
SAFEZA deleted all processed data and decommissioned the system after the pilot concluded. Discussions regarding a permanent deployment remain ongoing between the firm and the institution. A full rollout would require new students and employees to enrol their facial vectors into the system. Expanded features under consideration include live weapon detection for firearms, number plate recognition, and driver-vehicle association.
How the University of the Western Cape pilot operated?
How long did the facial recognition pilot run at the University of the Western Cape?
The pilot ran for five consecutive calendar days, spanning three working days and one weekend.
Which companies provided the technology for the campus trial?
Swiss-South African firm SAFEZA provided the technology, while Johannesburg-based Gigabrain supplied the generative AI inference engine.
What happens to the data collected during the test phase?
SAFEZA stated that all processed data was completely deleted and the system was decommissioned after the pilot concluded.
What additional security features could a full deployment include?
A full deployment could include live weapon detections for firearms, number plate recognition, and driver-vehicle association.
Related reading