Presense Detecion and counting ACAP
Detects how many objects are within the cameras scene and fires event on presence and count transition. Users can select to detect heads/faces, Humans/body or vehicles. Targets non-critial use cases for limited objects in scens, like
- Automation
- Queue monitoring
- Dwell-time
Download Presence Detection
Features
- Easy configuration with visual validation
- Filter detection area
- Adjust event timining depending on use-case
Configuration
- Set detection area - Click the buton and use the mouse to define the area the heads must be present
- Set ignore area - Use this filter if there is a small area in the area-of-interest that should be excluded.
- Minimum head size - Use this filter to eliminate small false detections.
- Maximum head size - Use this filter to eliminate large false detections.
- Confidence level - Adjust depending on the false detections. Start with Very low, view the augmentation and move up level if needed.
- Trigger after - Defines how long a head detection needs to be stable to impact the counter. This reduces false detections for sporadic changes. Use with caution as this may also result in missing detections for scenes where head detection is unstable. Note that the detection may lose tracking of a a specific head detection. A new detection may occur on the same head having a new timer starting
- Trigger timeout - After a positive presence state it will be kept high until timeout. New detections during timeout will extend timeout timer.
- Minimum presence - For a use case that is only interested when more than 1 person is detected in the scene.
Integration
Two events are fired:
- A counter value fires on count transition changes (a person leavs or enters)
- Presence true/false if there are anybody present.
ONVIF Events
For clients supporting ONVIF, use the event stream.
MQTT
Use the Camera MQTT client and setup a publish on event “Precence counter” or “Presence detection”.
You can also use SIMQTT that automaitically publish events.
Topic
simqtt/acap/presence/counter
Payload
{
"name": "If defined in SIMQTT",
"location": "If defined in SIMQTT",
"device": "ACCCxxxxxxxx",
"timestamp": 1676099193751,
"value": 1,
"event": "acap/presence/counter"
}
Topic
simqtt/acap/presence/presence/[true/false]
Payload
{
"name": "If defined in SIMQTT",
"location": "If defined in SIMQTT",
"device": "ACCCxxxxxxxx",
"timestamp": 1676099193751,
"state": true,
"event": "acap/presence/presence"
}
Things to consider
- Axis camera needs to support AOA (Axis Object Analytics)
- Detection quality depends on scene, camera placement and light conditions