Programming Routines and Tasks
This section contains reference information for application programming interface (API) routines and tasks associated with ENVI Deep Learning.
Routines
The following routines are available for use in writing custom tasks and workflows in an ENVI+IDL desktop environment. You cannot use routines in an enterprise environment such as the Geospatial Services Framework (GSF). In that case, use the Tasks listed below, or write a custom task that calls one or more of these routines.
- ENVIBoundingBoxSet
- ENVIDeepLearningGeoJSONToROI
- ENVIDeepLearningLabelRaster
- ENVIDeepLearningObjectDetectionRaster
- ENVIDeepLearningRaster
- ENVIDeepLearningROIToGeoJSON
- ENVITensorBoard
- ENVITensorFlowModel
- ENVITensorFlowObjectModel
Tasks
The following ENVITasks are available:
- BuildDeepLearningRaster
- BuildLabelRasterFromClassification
- BuildLabelRasterFromROI
- BuildObjectDetectionRasterFromAnnotation
- BuildObjectDetectionRasterFromROI
- BuildObjectDetectionRasterFromVector
- ClassActivationToClassification
- ClassActivationToPixelROI
- ClassActivationToPolygonROI
- ClassActivationToPolygonShapefile
- ClassActivationToPolylineShapefile
- ExtractDeepLearningLabelRasterFromFile
- ExtractDeepLearningRasterFromFile
- ExtractTensorFlowModelFromFile
- InitializeENVINet5MultiModel
- PostProcessObjectClassification
- RandomizeTrainTensorFlowMaskModel
- TensorFlowMaskClassification
- TensorFlowObjectClassification
- TrainTensorFlowMaskModel
- TrainTensorFlowObjectModel
Various parameter classes are also available for each data type in ENVI Deep Learning.