A raw model detection is not the same as an operational event. BAV ONE uses Vision Policies to evaluate detections against business context before deciding whether to create an event, generate an alert or trigger a workflow.
What a Vision Policy evaluates
A Vision Policy can combine object class, confidence threshold, configured zone or line, movement direction, time schedule, dwell duration, severity level and accountable response workflow. Only when the configured conditions are met does the policy create an event.
Why this matters
Without a policy layer, every detection would generate an alert. This creates alert fatigue and reduces operational value. Vision Policies allow teams to define exactly what matters in their specific operating environment.
An example policy
A restricted-zone policy might specify: when a person is detected inside a configured polygon zone, during after-hours schedule, for more than 30 seconds, create an event with severity high, attach permitted evidence and notify the security team.
Policies are site-specific
The same camera scene may require different policies for different shifts, zones or object classes. Vision Policies are configured per workflow rather than applied globally.
Detection vs decision
BAV ONE separates the detection layer (what the model sees) from the decision layer (what the policy determines). This separation makes the system more accountable and easier to validate.