Video analytics planning continues after installation: review unwanted alerts, investigate missed events and document each configuration change against the intended operating task.
Choose a specific event
First, describe what the business wants to review and where it occurs. Distinguish normal activity from the event that needs attention. Also specify the operating hours. A clear example gives the provider a stronger test brief than a general request for intelligent cameras.
Review the sample alerts
Ask the provider to demonstrate representative activity with permission from the property team. Include routine movement as well as the agreed test event. Then review the results with the intended operators. Record unhelpful notifications and missed test events for the provider to investigate.
Connect context to review
Identify the door and alarm records that help staff understand each event. Ask how a reviewer reaches the matching camera view. Also confirm who receives the notification and what they should do next. Check the proposed integration directly because product and service capabilities vary.
Keep tuning accountable
Finally, assign a person to review alert quality after handover. Keep a short record of changes and their reasons. Test again when the camera view, schedule or site activity changes. Measure whether the alert helps the operator make the intended decision.
Separate alert quality from review workload
Track how many alerts operators receive, how many are relevant and how long review takes. Also investigate known events that produced no alert. A quieter inbox can result from a better scene definition, but it can also result from missing activity. Both sides of the record matter.
For an illustrative review period, 30 relevant alerts out of 50 reviewed alerts gives 60% relevance for that sample. It does not reveal how many relevant events occurred without an alert. Compare the log with agreed observations or staged tests before concluding that a tuning change helped.
Use a controlled change log
- Record the camera, application version, scene and current behavior.
- Keep examples of the unwanted or missed event.
- Have the authorized person make one understood change.
- Repeat the original test and a routine-activity test.
- Record the result, remaining limits and rollback decision.
Check manufacturer requirements when the view, image size or application changes. The human-viewing pixel references do not establish analytics accuracy. For a new application rather than routine tuning, use the structured analytics trial guide to define the pilot first.
Continue learning
Explore the camera planning guides and manufacturer and platform comparison.
Professional help from NERSA
VideoSurveillancePA provides educational guidance. For property-specific help, visit NERSA’s commercial video surveillance installation.