Channels Configuration Parameters

The parameters in this section configure the number of processing channels Media Server requests from your License Server.

OpenText recommends that you configure the number of processing channels if you have multiple Media Server instances (to prevent a single server from consuming all of the available channels).

If an action requires a specific type of channel but insufficient channels are available, the action fails.

Channel Type Permitted Operations Allocating Channels
VisualChannels
  • Ingesting records from another Media Server for further processing.
  • All analysis operations (except audio processing).
  • Face recognition on large databases (maximum 250,000 faces, every additional 250,000 faces requires an additional visual channel).

    NOTE: The FaceClustering action is limited to 250,000 images per channel (not 250,000 face identities).

  • All surveillance and video management features.

The only operations not included in a visual channel are those for audio processing.

Allocate one channel for each concurrent process action that includes any visual analytics.

For example, to run OCR and object recognition on one video stream, set VisualChannels=1.

SurveillanceChannels
  • Face detection
  • Number plate recognition (ANPR)
  • Object class recognition (but only with certain pre-trained recognizers).
  • Scene analysis
  • Text detection

The following features are included if they are used with face detection, ANPR, scene analysis, or text detection:

  • Clothing analysis
  • Color clustering
  • Face recognition on standard databases (maximum 250 faces, every additional 250 faces requires an additional surveillance channel).

    NOTE: The FaceClustering action is limited to 250 images per channel (not 250 face identities).

  • Face demographics
  • Face state analysis
  • Vehicle make/model analysis

A surveillance channel also includes all video management features.

Allocate one channel for each analysis task, multiplied by the number of concurrent process actions. For example, to run number plate recognition and scene analysis on the video from two cameras set SurveillanceChannels=4.

To run face detection and face recognition on the video from one camera (with a database of up to 250 faces), you can set SurveillanceChannels=1, because face recognition is included in the surveillance pack when face detection is used.

AudioChannels
  • Audio categorization
  • Audio matching
  • Language identification
  • Speaker identification
  • Speech-to-text
  • All video management features.
Allocate one channel for each concurrent process action that includes audio processing. For example, if you want to run object recognition, language identification, and speech-to-text on one video stream, set VisualChannels=1 and AudioChannels=1.
VideoManagementChannels
  • Ingest (apart from the Receive ingest engine, which requires a visual channel).
  • Analysis:

    • Keyframe
    • News segmentation
    • Text segmentation
  • Encoding
  • Event stream processing
  • Transformation
  • Output

Allocate one video management channel for each concurrent process action that does not require a visual, surveillance, or audio channel.

Visual, surveillance, and audio channels all include these operations, so you do not need to allocate a video management channel for any action that is already allocated another type of channel.