Bodies
Body video analytics in OMNI Agent is available through the following prebuilt scenarios:
Bodies: Perimeter and Work Zone Monitoring (Overview / Top View)
OMNI Agent detects and tracks the body and key skeletal points: nose, left ear, left shoulder, left elbow, left wrist, left hand center, left hip, left knee, left ankle, right ear, right shoulder, right elbow, right wrist, right hand center, right hip, right knee, right ankle.Bodies: Behavioral Analytics (Overview / Top View)
OMNI Agent can detect and track human actions in video streams: sitting and lying (top view), falling, fighting, sitting and lying (overview).
Scenario settings
To change scenario settings, click the ✏️ icon next to the added scenario.
note
All prebuilt scenarios include the following groups of settings:
- ROI/Line Settings — manage regions of interest and crossing lines
- Output Settings — enable anonymous mode
- Event Settings — configure triggering logic and event delivery
This section covers only scenario-specific settings.
Perimeter and work zone monitoring
| Setting | Description |
|---|---|
| Video Analytics Zone | Allows configuring the area of the frame to process the video stream. |
| Use CUDA for Detector | Toggle on to enable GPU acceleration. Requires Nvidia GPU with CUDA and cuDNN installed. |
| Enable TensorRT | Toggle on to further accelerate processing with optimized models. Requires TensorRT SDK, CUDA, cuDNN, and Nvidia GPU. |
| Minimum Tracking Confidence | Default: 0.300. Minimum confidence required to continue tracking an already detected object. |
| Minimum Detection Confidence | Default: 0.100. Objects detected below this threshold are ignored. |
| Tracking Start Confidence | Default: 0.610. Detector confidence threshold for starting to track an object. |
| Keypoint Detection Confidence | Default: 0.500. Skeletal keypoints below this confidence threshold are ignored. |
| Activation Group ID | Used to start/stop video analytics scenarios via API. |
Behavioral analytics
| Setting | Description |
|---|---|
| Lie Detection Confidence | Default: 0.800 |
| Fall Detection Confidence | Default: 0.800 |
| Fight Detection Confidence | Default: 0.820 |
| Sitting Detection Confidence | Default: 0.750 |
Behavioral event settings
| Event Setting | Description |
|---|---|
| Enable Sitting Events | Repeat event skip interval for sitting events, ms. Default: 5000. Repeat events of this type will not be sent during this interval. |
| Enable Lying Events | Repeat event skip interval for lying events, ms. Default: 5000. Repeat events of this type will not be sent during this interval. |
| Enable Fall Events | Repeat event skip interval for fall events, ms. Default: 5000. Repeat events of this type will not be sent during this interval. |
| Enable Fight Events | Repeat event skip interval for fight events, ms. Default: 5000. Repeat events of this type will not be sent during this interval. |