Processor Settings
The following tables list the Kraken Processor settings.
The Markup and Metadata settings are common across different processors, while other settings vary depending on the selected Processor Module.
| Processor Setting | Default | Description/Values |
|---|---|---|
| Enable VMTI KLV Generation | Enabled | Check this checkbox to enable Video Moving Target Indicator (VMTI) KLV generation. When enabled, this tracks the MISB ST 0903 VMTI KLV data (carried in MISB ST0601 metadata packets) reported by the processor in use. Some common data points include:
The above list is only a sample of data points that may be available. |
| Draw Object Detection Boxes | Enabled | Check this checkbox to view object detection boxes on your stream. |
| Draw Labels for Detected Objects | Enabled | Check this checkbox to show detected object labels (e.g. car, boat, person). |
| Object Detection Box Color | Blue | Select the object detection box color from the dropdown menu. Color options are:
|
| Border Size for Object Detection Boxes | 1 | To modify the border size of the object detection boxes, type in a value here. Range is from -16 to 16 pixels |
Shield AI Tracker Settings | Default | Description/Values |
|---|---|---|
| Type of analytics algorithm to run | LAND | The Shield AI Tracker allows you to specify an algorithm suited to certain environments. Environment options are:
|
| Sensitivity of the Tracker analytic | Dependent on Analytic | Sensitivity values for the Shield AI Tracker Processor module are related to the chosen analytic. For all analytics, a value of 0 will generate no detections. For the LAND analytic (Default 5): For the Maritime Analytic (Default 10): For the Search and Rescue (SAR) Analytic (Default 10): |
| Rectangular area around the border of the image to ignore | 100 | Input a number between 0 and 500 to adjust the amount of space that will be ignored around the detected object. |
| Yolo-based image processing Settings | Default | Description/Values |
|---|---|---|
| Minimum Confidence Threshold for detections. | 0.6 | Input a number between 0.01 and 1 to set the minimum confidence threshold for detections. Higher number will result in fewer detections, but will improve accuracy by reducing false positives. |
Intersection over Union (IoU) threshold for non-maximum suppression. | 0.45 | Input a number between 0.01 and 1 to set the IoU threshold for non-maximum suppression. The Intersection over Union value is determined by dividing the overlapping area between two detection boxes (the Area of Intersection), by the total area covered by those detection boxes (the Area of Union). As long as both boxes are seeking the same type of object, a high Area of Intersection relative to the Area of Union means that the boxes are likely to be detecting the same object. The value input in this field determines the percentage of overlap between two bounding boxes that is deemed high enough to rule-out multiple detections. Higher numbers filter out more redundant boxes. |
| Maximum number of detections per frame. | 30 | Input a number between 1 and 100 to set of maximum number of detections that can be made per frame. |
| Type of objects to detect. | n/a | Type in the name of the object you would like the processor to detect. Values can be generic nouns such as 'car', 'person', or 'boat.' TBD limits to this input??? Can objects be specified with additional adjectives, e.g. 'red car' ? |
| Type of application | object-detector | Select the type of insight to be processed. Options are:
|