Bounding Box

Encapsulates the settings for common object detection. Common object detection provides a detection stage that precedes the execution of the EyesOnIt Large Vision Model. If common object detection is enabled, the Large Vision Model will only run if a common object is detected. If common object detection is disabled, the common object detection stage gets skipped and the Large Vision Model runs without a check for common objects.

Note that the Large Vision Model will detect common objects more accurately than the bounding box model in many cases. If you don't get the accuracy that you need with the bounding box model, considering setting the threshold value low. This will send anything that could be an object to the Large Vision Model for validation. If you still don't get the results you need, consider disabling common object detection and just relying on the Large Vision Model. The Large Vision Model will always run if the bounding box model is disabled.

Example

"bounding_box": {
	"bounding_box_enabled": true,
	"detect_people": false,
	"detect_vehicles": true,
	"detect_bags": false,
	"person_confidence_threshold": 30,
	"vehicle_confidence_threshold": 30,
	"bag_confidence_threshold": 30
}

Attributes

Name	Type	Description
bounding_box_enabled	boolean	true to enable detection of common objects, false to disable detection
detect_people	boolean	if true, the bounding box model will attempt to detect people. If false, the common object model will not attempt to detect people
detect_vehicles	boolean	if true, the bounding box model will attempt to detect vehicles. If false, the common object model will not attempt to detect vehicles
detect_bags	boolean	if true, the bounding box model will attempt to detect bags. If false, the common object model will not attempt to detect bags
person_confidence_threshold	int	the detection threshold for people from 10 to 100
vehicle_confidence_threshold	int	the detection threshold for vehicles from 10 to 100
bag_confidence_threshold	int	the detection threshold for bags from 10 to 100