Create a Session Configuration
To create a configuration, you need to consider:
- the media source, because this determines how to ingest the video.
- which analysis tasks you want to run, for example face recognition or speech-to-text.
- the way in which data will flow between tasks. For example, you cannot run face recognition without first running face detection. You must configure the input of the face recognition task to be the output from the face detection task.
- how to output data.
A configuration must have a section named [Session]
. This section contains a list of tasks that you want to run:
[Session] Engine0=Ingest Engine1=FaceDetect Engine2=FaceRecognize Engine3=EncodeImages Engine4=OutputToIDOL ...
NOTE: Task names cannot include any of the following characters:
- period (
.
) - comma (
,
) - colon (
:
) - semicolon (
;
) - asterisk (
*
) - equals sign (
=
)
These tasks are configured in other sections of the configuration file. The previous example defines a task named Ingest
which you would configure in a section named [Ingest]
.
Every task section must include the Type
parameter to specify the engine to use to complete the task, and any settings that the engine requires to complete the task. For example:
[Ingest] Type=Video