Object Recognition

Object Recognition recognizes known objects that appear in media. Object recognition identifies specific objects, not classes of objects.

Before using object recognition, you must train Media Server by adding training images to the training database. For instructions that describe how to train Media Server, refer to the Media Server Administration Guide.

Configuration Parameter Description
ColorAnalysis Whether to perform color analysis on detected objects to reduce false detections.
Database An object database to use to recognize objects in an image.
Geometry The analysis mode to use when comparing the source media to 2-D objects in the training database.
Geometry3 The analysis mode to use when comparing the source media to 3-D objects in the training database.
Identifier A comma-separated list of identifiers that specifies the objects to match against.
Input The image track to process.
MaxInputQueueLength Can be used to place a limit on latency.
MaxSize The maximum expected size of an object.
Metadata A metadata field name and value to specify the objects to recognize.
MinSize The minimum expected size of an object.
NumParallel The maximum number of video frames to analyze simultaneously.
ObjectEnvironment The type of background that a partially or completely transparent object is superimposed on.
Occlusion Whether the object might be partially hidden in images.
Region A region of the frame to restrict processing to.
RegionUnit The units that the Region parameter uses to specify the size and position of a region.
RestrictToInputRegion Specifies whether to analyze a region of the input image or video frame that is specified in the input record, instead of the entire image.
SampleInterval The interval at which frames are selected to be analyzed.
SegmentDuration The maximum duration of a record in the SegmentedResult or SegmentedResultWithSource track.
SizeUnit The units to use for setting the MinSize and MaxSize parameters.
SyncDatabase Specifies whether to synchronize with the training database before beginning the analysis task.
Type The analysis engine to use. Set this parameter to ObjectRecognition.

Output Tracks

Output track Type Description Output1This column indicates whether the information contained in the track is included by default in the output created by an output task (when you don't set the Input parameter for the output task).
Data ObjectRecognitionResult Contains one record, describing the analysis results, per recognized object, per video frame. No
DataWithSource ObjectRecognitionResultAndImage

The same as the Data track, but each record also includes the source frame.

No
Event ObjectRecognitionEvent (Deprecated) Media Server creates a record in the Event track when an object appears or disappears. Yes
Result ObjectRecognitionResult Contains one record, describing the analysis results, for each recognized object. When the same object appears in many consecutive frames, Media Server produces a single result. Yes
ResultWithSource ObjectRecognitionResultAndImage

The same as the Result track, but each record also includes the best source frame.

No
SegmentedResult ObjectRecognitionResult (Available only when processing video). Contains one or more records, describing the analysis results, for each recognized object. The maximum duration of records in this track is limited by the parameter SegmentDuration, so multiple records are created with the same ID when an object remains in the scene for longer than this duration. Yes
SegmentedResultWithSource ObjectRecognitionResultAndImage

(Available only when processing video). The same as the SegmentedResult track, but each record also includes the best source frame.

No
Start ObjectRecognitionResult

The same as the Data track, except it contains only the first record of each event.

No
End ObjectRecognitionResult

The same as the Data track, except it contains only the last record of each event.

No

ObjectRecognitionResult

Field name Type Description
id UUIDData A universally unique identifier to identify the object
identity IdentityData The identity of the model in the database that matches the detected object
boundary PolygonData A polygon to indicate the location of the object in the frame

ObjectRecognitionResultAndImage

The same as ObjectRecognitionResult records, with the following additional fields.

Field name Type Description
image ImageData The source frame

_FT_HTML5_bannerTitle.htm