文档

mxpi_motsimplesortV2

Function

Implements multi-object tracking (MOT). The differences between this version and the previous version are as follows:

  • In the V2 version, the input port of the plugin is adjusted. When only the object bounding box information is used for MOT, only one input port needs to be connected. On this basis, if the characteristics of the object are further used for MOT, two input ports are connected.
  • dataSource implements automatic configuration.

Synchronous/Asynchronous (Status)

Synchronous

Constraints

None

Plugin Base Class (Factory)

mxpi_motsimplesortV2

Input/Output

Input: buffer (data type: MxpiBuffer) and metadata (data type: MxpiObjectList)

Output: buffer (data type: MxpiBuffer) and metadata (data type: MxpiTrackLetList)

Port Format (Caps)

Static input: {"ANY"}; dynamic input: {"ANY"}

Static output: {"ANY"}

Property

For details, see Table 1.
Table 1 mxpi_motsimplesortV2 plugin properties

Property Name

Description

Mandatory or Not

Modifiable or Not

dataSourceDetection

Index of the object bounding box data after model detection. The default value is the key value of the output port of the upstream plugin.

No

Yes

dataSourceFeature

Index of the object feature data after feature extraction. The default value is the key value of the output port of the upstream plugin.

No

Yes

trackThreshold

Threshold for the probability that the traced objects are the same one. If the probability is greater than the threshold, the traced objects are the same one. The value ranges from 0 to 1, and defaults to 0.5.

No

Yes

lostThreshold

Threshold for the number of lost frames of a traced object. If the number of lost frames is greater than the threshold, the traced object is lost. The value ranges from 0 to 10, and defaults to 5.

No

Yes

Example

  • Only the bounding box information of an object is used for MOT. Typical scenario: tracking the vehicles in a video.

  • Both the bounding box information and features of an object is used for MOT. Typical scenario: identifying the locations of the people or objects in a video.