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 ...

D. Implemented processors in the streams framework

Processors in Package stream.coffee.helper

This package provides several processors useful for processing the coffee dataset.

Processor AddCoffeeLabels

This processor adds the true label to the video frames of one coffee capsule video file.

One label is attached to each data item: @label:event, telling, the color of the coffee

capsule slipping down the slide or ”no event”, in case no capsule is shown within the

frame (multinominal).

The labels have to be stored in a file, that gets read in before the stream process starts

(during the init of this processor).

Parameter Type Description Required

file String Sets the file the labels are stored in. false

Table .17.: Parameters of class AddCoffeeLabels.

Processor DatasetGenerator

This processor creates a dataset, containing only one data item for each coffee capsule

event. If the last data item belonged to an event, but the current does not, the minimal

RGB values for that event plus the label are returned. All other data items are simply

dropped. The minimal color values observed during the event have to be included in

the data item already. This can be done using the

stream.coffee.tagging.MaxRGBOverEvent processor.

116

More magazines by this user
Similar magazines