WEBVTT

00:00:00.000 --> 00:00:01.400 position:50% size:92% align:center
In Config Tool, open Description and enable natural-language detection.

00:00:01.400 --> 00:00:03.200 position:50% size:92% align:center
Double-click the blank description row to define the alerting target.

00:00:03.200 --> 00:00:06.600 position:50% size:92% align:center
Describe one target: person wearing a yellow top and dark pants.

00:00:06.600 --> 00:00:07.800 position:50% size:92% align:center
Keep Can alert on and Background Description off. Set the match threshold.

00:00:07.800 --> 00:00:09.800 position:50% size:92% align:center
This test uses 90. Tune the threshold against representative footage.

00:00:09.800 --> 00:00:11.200 position:50% size:92% align:center
Save the target description.

00:00:11.200 --> 00:00:14.400 position:50% size:92% align:center
Add a broad background: person wearing everyday clothing. Can alert stays off.

00:00:14.400 --> 00:00:15.800 position:50% size:92% align:center
Save the broad background. New people can have any other appearance.

00:00:15.800 --> 00:00:17.800 position:50% size:92% align:center
Check that the detection condition requires at least one matching object.

00:00:17.800 --> 00:00:21.400 position:50% size:92% align:center
Open Alert Rules and add a Detection condition rule.

00:00:21.400 --> 00:00:22.800 position:50% size:92% align:center
Name the rule Person wearing a yellow top.

00:00:22.800 --> 00:00:25.200 position:50% size:92% align:center
Use 1 s dwell and 5 s reset. Choose severity and active hours for your procedure.

00:00:25.200 --> 00:00:28.000 position:50% size:92% align:center
Save the rule, then apply the complete camera configuration.

00:00:28.000 --> 00:00:30.400 position:50% size:92% align:center
Start analytics and wait for Running.

00:00:30.400 --> 00:00:32.200 position:50% size:92% align:center
Return to Description and check the live preview.

00:00:32.200 --> 00:00:40.200 position:50% size:92% align:center
Check the highlighted yellow-top person against the surrounding activity.
