PostProcessObjectClassification Task

This task refines an object classification shapefile produced by the DeepLearningObjectClassification task. If the initial result does not accurately identify features, this task lets you test and preview different parameter values. When the previewed result looks better, you can then output a new shapefile.

Example

Sample data files are available on our ENVI Tutorials web page. Click the "Deep Learning" link in the ENVI Tutorial Data section to download a .zip file containing the data. Extract the contents of the .zip file to a local directory. Files are located in the object_detection folder.

This example classifies an image using a trained object detection model that looks for parking spots with handicap signs painted on the pavement. It displays the initial result, then invokes the PostProcessObjectClassification task. It applies different CONFIDENCE_THRESHOLD and IOU_THRESHOLD values to improve the classification. Finally, it creates a new output shapefile.

; Start the application

e = ENVI()

; Select a raster to classify.

; Update the following line with the correct path

; to the tutorial data files.

RasterFile = 'C:\MyTutorialFiles\ImageToClassify.dat'

Raster = e.OpenRaster(RasterFile)

; Select a trained model.

; Update the following line with the correct path

; to the tutorial data files.

ModelFile = 'C:\MyTutorialFiles\ObjectDetectionModel_HandicapSpots.envi.onnx'

Model = ENVIDeepLearningObjectModel(ModelFile)

; Perform an initial classification

ClassTask = ENVITask('DeepLearningObjectClassification')

ClassTask.INPUT_RASTER = Raster

ClassTask.INPUT_MODEL = Model

ClassTask.CONFIDENCE_THRESHOLD = 0.4

ClassTask.IOU_THRESHOLD = 0.7

ClassTask.Execute

; Display the initial result

View = e.GetView()

Layer = View.CreateLayer(Raster)

Layer2 = View.CreateLayer(ClassTask.OUTPUT_VECTOR)

View.Zoom, /FULL_EXTENT

Next, zoom into an area with overlapping bounding boxes over the same feature. Enter these commands at the ENVI prompt in the IDL Console:

View.GoToLocation, 3114967.3777D,1765077.1140D, /MAP

View.Zoom, '400%', LAYER=Layer

Invoke the PostProcessObjectClassification task with new threshold values:

Task = ENVITask('PostProcessObjectClassification')

Task.INPUT_VECTOR = ClassTask.OUTPUT_VECTOR

Task.CONFIDENCE_THRESHOLD = 0.8

Task.IOU_THRESHOLD = 0.1

Task.OUTPUT_VECTOR_URI=e.GetTemporaryFilename('shp')

Task.Execute

Hide the initial shapefile and display the post-processed shapefile:

Layer2.Close

Layer3 = ViewCreateLayer(Task.OUTPUT_VECTOR)

Syntax

Result = ENVITask('PostProcessObjectClassification')

Input parameters (Set, Get): CONFIDENCE_THRESHOLD, INPUT_VECTOR, IOS_FILTER, IOS_THRESHOLD, IOU_THRESHOLD, OUTPUT_VECTOR_URI

Output parameters (Get only): OUTPUT_VECTOR

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

CONFIDENCE_THRESHOLD (optional)

Specify a floating-point threshold value between 0 and 1.0. Bounding boxes with a confidence score less than this value will be discarded before applying IOU_THRESHOLD. The default value is 0.2. Decreasing this value generally results in more classification bounding boxes throughout the scene. Increasing it results in fewer classification bounding boxes.

INPUT_VECTOR (required)

Specify the classification vector to be processed.

IOS_FILTER (optional)

Set this parameter to true to filter out bounding boxes that have an Intersection over Self (IOS) value larger than IOS_THRESHOLD.

IOS_THRESHOLD (optional)

Specify a floating-point value between 0 and 1.0, indicating the Intersection over Self (IOS) value. This parameter fixes any overlapping bounding boxes that remain after applying the IOU_THRESHOLD, particularly smaller boxes inside of larger ones. The default value is 0.8.

IOU_THRESHOLD (optional)

Specify a floating-point value between 0 and 1.0, indicating the Non-Maximum Suppression Intersection over Union (NMS IOU) value. This is a Deep Learning object detection parameter that reduces detection clustering by pruning predicted bounding boxes that have high IOU overlap with previously selected boxes. The default value is 0.5. Increasing this value results in more bounding boxes around identified features. Decreasing the value results in fewer bounding boxes.

OUTPUT_VECTOR_URI (optional)

Specify a string with the fully-qualified path and filename for OUTPUT_VECTOR.

Output Parameters

OUTPUT_VECTOR

This is a reference to the output vector.

Methods

Execute

Parameter

ParameterNames

See ENVI Help for details on these ENVITask methods.

Properties

DESCRIPTION

DISPLAY_NAME

NAME

REVISION

Version History

Deep Learning 1.2

Introduced