Abstract Classes

Abstract classes for running the Distant Viewing Toolkit.

The objects VisualInput, FrameAnnotator, and Aggregator are abstract classes. They correspond to the method for extracting visual inputs from digitized files, methods for extracting metadata from visual images, and methods for aggregating the data across a collection. The toolkit provides many ready-made implementions of these classes in order to solve common tasks.

class dvt.abstract.VisualInput(**kwargs)

Bases: abc.ABC

Base class for producing batches of visual input.

Implementations in the toolkit provide inputs for video files and collections of still images. Users can further implement additional input types if needed.

get_metadata()

Provide metadata from the input connection.

Returns an object that can be processed with the process_output_values utility function. The output is stored as a phantom metadata annotator.

next_batch()

Move forward one batch and return the current FrameBatch object.

When called, this should return a FrameBatch object. It will be iteratively called until a batch is returned with the continue_read flag is set to false.

open_input()

Restart and initialize the input feed.

This method is called once before passing data through a collection of annotators.

class dvt.abstract.FrameAnnotator(**kwargs)

Bases: abc.ABC

Base class for annotating a batch of frames.

Subclasses of this abstract class take subsets of frames, composed as FrameBatch objects, and return annotated data. Several common annotations are implemented in the toolkit. Users can create their own annotations by implementing the __init__ and annotate methods. Note that the annotator must contain a name attribute; the name is used as the key in the output annotation object.

annotate(batch)

Annotate a batch of frames and return the resulting annotations.

This method contains the core functionality of an annotator. It takes a batch of frames and returns the annotated output in a format that can be processed by process_output_values.

Parameters:batch (FrameBatch) – A batch of images to annotate. Returns:The method should return an item that can be processed by the utility function process_output_values.

name = 'abstract'

class dvt.abstract.Aggregator(**kwargs)

Bases: abc.ABC

Base class for aggregating the output from a pipeline of processors.

Aggregators take all of the information contained in a set of annotators and produce new metadata. The aggregators do not have access to the original visual material.

name

A description of the aggregator. Used as a key in the output data.

Type:str

aggregate(ldframe, **kwargs)

Aggregate annotations.

Parameters:ldframe (dict) – A dictionary of data pased from a DataExtraction object. Returns:An object that can be processed by process_output_values.

name = 'abstract'