Facial Recognition vs Object Detection

One of the most commercially consequential misunderstandings in AI surveillance is conflating facial recognition with general-purpose AI camera analytics. These are fundamentally different technologies with different data outputs, different PDPL compliance requirements in Saudi Arabia, and different appropriate use cases.

What Object Detection Does (and Does Not Do)

Object detection neural networks classify objects in a video frame into categories — person, vehicle, hard hat, fire, crowd. They output bounding box coordinates and category labels. They do not identify individuals. No biometric template is created. No database is queried. The system cannot tell you who a person is — only what they are doing and where.

What Facial Recognition Does

Facial recognition extracts biometric templates from detected face regions — mathematical representations of facial geometry that uniquely identify individuals. These templates are compared against a database of known individuals to produce identity matches. Facial recognition creates a record linking a real person's identity to their presence at a specific location at a specific time — subject to the strictest data protection requirements under Saudi PDPL.

PDPL Implications of Each

Object detection analytics process anonymized movement data — no individual is identified. PDPL still applies to the footage itself, but analytics output contains no personal data. Facial recognition creates biometric personal data at the point of template extraction, requiring explicit consent or lawful basis under PDPL, visible disclosure notices, and strict retention and deletion obligations.