deepscribe.pipeline.analysis

class deepscribe.pipeline.analysis.AnalysisTask(*args, **kwargs)

Abstract class that requires the completion of dataset selection and model training.

requires()

Task depends on a trained model and a dataset archive.

Returns:Dictionary with TrainKerasModelFromDefinitionTask and SelectDatasetTask

class deepscribe.pipeline.analysis.GenerateClassificationReportTask(*args, **kwargs)

Generates a classification report from the trained model on test data.

output()

Location of the output image.

Returns:luigi.LocalTarget

run()

Loads the model, runs it on test data, saves the classification report as txt.

Returns:None

class deepscribe.pipeline.analysis.PlotConfusionMatrixTask(*args, **kwargs)

Produces a confusion matrix from the model on test data.

output()

Location of the output image.

Returns:luigi.LocalTarget

run()

Loads the model, runs it on test data, saves the confusion matrix as png.

Returns:None

class deepscribe.pipeline.analysis.PlotIncorrectTask(*args, **kwargs)

Loads 16 incorrectly classified images from test data and plots them in a grid.

output()

Location of the output image.

Returns:luigi.LocalTarget

run()

Loads model, runs on test data, picks 16 random incorrectly classified images.

Returns:

class deepscribe.pipeline.analysis.RunAnalysisOnTestDataTask(*args, **kwargs)

WrapperTask requiring the tasks GenerateClassificationReportTask, PlotConfusionMatrixTask, and PlotIncorrectTask.

requires()

Requires the tasks GenerateClassificationReportTask, PlotConfusionMatrixTask, and PlotIncorrectTask.

Returns:List of luigi.Task subclasses