RandomizeTrainDeepLearningPixelModel Task

This task generates a number of randomized sets of parameter values to train a pixel-based deep learning model.

Example

; Start the application

e = ENVI(/HEADLESS)

; Get the task from the catalog of ENVITasks

RandomizeModelTask = ENVITask('RandomizeTrainDeepLearningPixelModel')

; Define task inputs

RandomizeModelTask.ITERATIONS = 5

; Run the task

RandomizeModelTask.Execute

; Print the output values for each iteration

Print, 'Blur Distance: ', RandomizeModelTask.OUTPUT_BLUR_DISTANCE

Print, 'Class Weight: ', RandomizeModelTask.OUTPUT_CLASS_WEIGHT

Print, 'Loss Weight: ', RandomizeModelTask.OUTPUT_LOSS_WEIGHT

Print, 'Solid Distance: ', RandomizeModelTask.OUTPUT_SOLID_DISTANCE

Syntax

Result = ENVITask('RandomizeTrainDeepLearningPixelModel')

Input parameters (Set, Get): BLUR_DISTANCE, CLASS_WEIGHT, ITERATION, LOSS_WEIGHT, METHOD, SOLID_DISTANCE

Output parameters (Get only): OUTPUT_BLUR_DISTANCE, OUTPUT_CLASS_WEIGHT, OUTPUT_LOSS_WEIGHT, OUTPUT_SOLID_DISTANCE

Properties marked as "Set" are those that you can set to specific values. You can also retrieve their current values any time. Properties marked as "Get" are those whose values you can retrieve but not set.

Input Parameters

BLUR_DISTANCE (optional)

Specify a double-precision array with the minimum and maximum distance in pixels to expand positive (non-background) areas beyond the solid distance expansion. The area expanded is blurred, diminishing from the edges of the feature areas to the blur distance. The maximum blur distance is used at the beginning of training and decreased to the minimum blur distance at the end of training. If this value is not set, OUTPUT_BLUR_DISTANCE will be a random value within normal range.

CLASS_WEIGHT (optional)

Specify a double-precision array with the minimum and maximum weights for having a more even balance of classes (including background) during patch selection. Balancing of patch selection is weighted by the maximum value at the beginning of training and decreased to the minimum value at the end of training. In general, set the maximum higher for sparser training sets. The useful range for the maximum value is between 0.0 and 3.0. If this value is not set, OUTPUT_CLASS_WEIGHT will be a random value within normal range.

ITERATIONS (required)

Specify the number of sets of parameter values to generate.

LOSS_WEIGHT (optional)

Specify the weight of positive (non-background) pixels when calculating how well the model is fitting the training data. If this value is not set, OUTPUT_LOSS_WEIGHT will be a random value within normal range.

METHOD (optional)

Specify a string indicating the method used to randomize the values. The choices are:

Random Uniform: produces a uniform distribution of numbers
Sobol Sequence (default): produces a non-random list of numbers that does a better job exploring the range of possible values

SOLID_DISTANCE (optional)

Specify the distance in pixels to expand positive (non-background) areas. If this value is not set, OUTPUT_SOLID_DISTANCE will be a random value within normal range.

Output Parameters

OUTPUT_BLUR_DISTANCE

A double-precision array with the minimum and maximum distance in pixels to expand positive (non-background) areas beyond the solid distance expansion.

OUTPUT_CLASS_WEIGHT

The double-precision array with the minimum and maximum weights for having a more even balance of classes (including background) when sampling.

OUTPUT_LOSS_WEIGHT

The weight of positive (non-background) pixels when calculating how well the model is fitting the training data.

OUTPUT_SOLID_DISTANCE

The distance in pixels to expand positive (non-background) areas.

Methods

Execute

Parameter

ParameterNames

See ENVI Help for details on these ENVITask methods.

Properties

DESCRIPTION

DISPLAY_NAME

NAME

REVISION

Version History

Deep Learning 3.0

Renamed from RandomizeTrainTensorFlowMaskModel task.