Before running analysis on video, you must create a Media Server configuration file that instructs Media Server which analysis operations to perform. You can either:
ConfigDirectory
parameter, in the [Paths]
section of the Media Server configuration file. In this case set the VideoServerConfigurationName
parameter when you configure CFS, so that CFS sends the name of the configuration to Media Server when it runs the process
action to start processing.VideoServerConfigurationFileName
parameter when you configure CFS, so that CFS sends the configuration file to Media Server when it runs the process
action to start processing.CFS sends files to Media Server for analysis, so configure Media Server to ingest files:
[Ingest] IngestEngine=Files [Files] Type=LibAv
By default, Media Server ingests video at playback speed. A video that contains five minutes of content takes five minutes to ingest. If you have configured only fast analysis operations, consider setting the configuration parameter IngestRate=0
. This specifies that Media Server ingests video as fast as possible without dropping frames. Be aware that if you have configured resource-intensive analysis operations such as OCR, ingestion with IngestRate=0
will be slower than playback speed, because in this mode Media Server cannot drop frames.
[Ingest] IngestRate=0 IngestEngine=Files [Files] Type=LibAv
For more information about the ingest rate, refer to the Media Server Administration Guide.
Create a section in the configuration file named [Analysis]
, and configure the analysis operations that you want to run.
The following example configures keyframe analysis and optical character recognition:
[Analysis] AnalysisEngine0=Keyframe AnalysisEngine1=OCR [Keyframe] Type=Keyframe KeyAtLeastSec=30 [OCR] Type=OCR FrameRate=2
For more information about configuring analysis in Media Server, refer to the Media Server Administration Guide.
CFS expects Media Server to return the results of analysis in the process
action response. You must create a section in the configuration file named [Output]
, and configure an output task to write data to the action response.
[Output] OutputEngine0=response
[response] Type=response
|