Real-time feature extraction from video stream data for stream ...

ai.cs.uni.dortmund.de

Real-time feature extraction from video stream data for stream ...

Processors in Package stream.coffee.eventdetection

This package provides processors to detect events and evaluate event detection

classifiers.

Processor EventDetectionEvaluation

This processor evaluates the quality of a learner, which classified frames as ”events” or

”no events”. As an event is said to be recognized, as soon as at least one frame of the

event was classified as ”event”, the evaluation processors, included in the core streams

module, are not sufficient to perform the evaluation.

Parameter Type Description Required

label String The key, under which the t label is stored. false

prediction String The key, under which the predicted label is stored. false

Table .18.: Parameters of class EventDetectionEvaluation.

Processor ThresholdEventDetection

This processor classifies the incoming frames as ”events” or ”no events”, based on the

value of one color channel. If the average color value of the color channel falls below a

given threshold t, the frame is labeled as an ”event” frame.

Parameter Type Description Required

attribute String Tells the processor, on which attribute to base the event

detection on.

t Integer Sets the threshold t to a new value. false

predictionkey String Sets the key under which the classifier stores the predicted

label.

standardvalue Integer Sets the value the attribute has in random frames. false

Table .19.: Parameters of class ThresholdEventDetection.

false

false

117

More magazines by this user
Similar magazines