ObjectTrackingProcessor Class

Object tracker processor allows for continuous tracking of one of more objects over a finite sequence of video frames. It must be used downstream of an object detector extension node, thus allowing for the extension to be configured to to perform inferences on sparse frames through the use of the 'maximumSamplesPerSecond' sampling property. The object tracker node will then track the detected objects over the frames in which the detector is not invoked resulting on a smother tracking of detected objects across the continuum of video frames. The tracker will stop tracking objects which are not subsequently detected by the upstream detector on the subsequent detections.

All required parameters must be populated in order to send to Azure.

Inheritance

azure.media.videoanalyzeredge._generated.models._models_py3.ProcessorNodeBase

ObjectTrackingProcessor

Constructor

ObjectTrackingProcessor(*, name: str, inputs: List[azure.media.videoanalyzeredge._generated.models._models_py3.NodeInput], accuracy: Optional[Union[str, azure.media.videoanalyzeredge._generated.models._azure_video_analyzerfor_edge_enums.ObjectTrackingAccuracy]] = None, **kwargs)

Variables

type

str

Required. Type discriminator for the derived types.Constant filled by server.

name

str

Required. Node name. Must be unique within the topology.

inputs

list[<xref:azure.media.videoanalyzer.edge.models.NodeInput>]

Required. An array of upstream node references within the topology to be used as inputs for this node.

accuracy

str or <xref:azure.media.videoanalyzer.edge.models.ObjectTrackingAccuracy>

Object tracker accuracy: low, medium, high. Higher accuracy leads to higher CPU consumption in average. Possible values include: "low", "medium", "high".