Tool Artificial Neural Network Classification

Integration of the OpenCV Machine Learning library for Artificial Neural Network classification of gridded features.

References

Author: O.Conrad (c) 2016, L.Piras (c) 2013
Menu: Imagery|Classification|Machine Learning

Parameters

	Name	Type	Identifier	Description	Constraints
Input	Features	Grid list, input	`FEATURES`	-	-
	Training Samples	Table, input	`TRAIN_SAMPLES`	Provide a class identifier in the first field followed by sample data corresponding to the input feature grids.	-
	Training Areas	Shapes, input	`TRAIN_AREAS`	-	-
Output	Classification	Grid, output	`CLASSES`	-	-
Output	Look-up Table (*)	Table, output, optional	`CLASSES_LUT`	A reference list of the grid values that have been assigned to the training classes.	-
Options	Grid System	Grid system	`PARAMETERS_GRID_SYSTEM`	-	-
	Normalize	Boolean	`NORMALIZE`	-	Default: 0
	Colors from Features	Boolean, GUI	`RGB_COLORS`	Use the first three features in list to obtain blue, green, red components for class colour in look-up table.	Default: 0
	Training	Choice	`MODEL_TRAIN`	-	Available Choices: [0] training areas [1] training samples [2] load from file Default: 0
	Class Identifier	Table field	`TRAIN_CLASS`	-	-
	Buffer Size	Floating point	`TRAIN_BUFFER`	For non-polygon type training areas, creates a buffer with a diameter of specified size.	Minimum: 0.000000 Default: 1.000000
	Load Model	File path	`MODEL_LOAD`	Use a model previously stored to file.	-
	Save Model	File path	`MODEL_SAVE`	Stores model to file to be used for subsequent classifications instead of training areas.	-
	Number of Layers	Integer	`ANN_LAYERS`	You can specify the number of layers in the network (not including input and output layer).	Minimum: 1 Default: 3
	Number of Neurons	Integer	`ANN_NEURONS`	You can specify the number of neurons in each layer of the network.	Minimum: 1 Default: 5
	Maximum Number of Iterations	Integer	`ANN_MAXITER`	-	Minimum: 1 Default: 300
	Error Change (Epsilon)	Floating point	`ANN_EPSILON`	Termination criteria of the training algorithm. You can specify how much the error could change between the iterations to make the algorithm continue (epsilon).	Minimum: 0.000000 Default: 0.000000
	Activation Function	Choice	`ANN_ACTIVATION`	-	Available Choices: [0] Identity [1] Sigmoid [2] Gaussian Default: 1
	Function's Alpha	Floating point	`ANN_ACT_ALPHA`	-	Default: 1.000000
	Function's Beta	Floating point	`ANN_ACT_BETA`	-	Default: 1.000000
	Training Method	Choice	`ANN_PROPAGATION`	-	Available Choices: [0] resilient propagation [1] back propagation Default: 1
	Initial Update Value	Floating point	`ANN_RP_DW0`	-	Default: 0.000000
	Increase Factor	Floating point	`ANN_RP_DW_PLUS`	-	Minimum: 1.010000 Default: 1.200000
	Decrease Factor	Floating point	`ANN_RP_DW_MINUS`	-	Minimum: 0.010000 Maximum: 0.990000 Default: 0.500000
	Lower Value Update Limit	Floating point	`ANN_RP_DW_MIN`	-	Minimum: 0.010000 Default: 0.100000
	Upper Value Update Limit	Floating point	`ANN_RP_DW_MAX`	-	Minimum: 1.010000 Default: 1.100000
	Weight Gradient Term	Floating point	`ANN_BP_DW`	-	Default: 0.100000
	Moment Term	Floating point	`ANN_BP_MOMENT`	-	Default: 0.100000
(*) optional

Command-line

Usage: saga_cmd imagery_opencv 11 [-FEATURES <str>] [-NORMALIZE <str>] [-CLASSES <str>] [-CLASSES_LUT <str>] [-MODEL_TRAIN <str>] [-TRAIN_SAMPLES <str>] [-TRAIN_AREAS <str>] [-TRAIN_CLASS <str>] [-TRAIN_BUFFER <double>] [-MODEL_LOAD <str>] [-MODEL_SAVE <str>] [-ANN_LAYERS <num>] [-ANN_NEURONS <num>] [-ANN_MAXITER <num>] [-ANN_EPSILON <double>] [-ANN_ACTIVATION <str>] [-ANN_ACT_ALPHA <double>] [-ANN_ACT_BETA <double>] [-ANN_PROPAGATION <str>] [-ANN_RP_DW0 <double>] [-ANN_RP_DW_PLUS <double>] [-ANN_RP_DW_MINUS <double>] [-ANN_RP_DW_MIN <double>] [-ANN_RP_DW_MAX <double>] [-ANN_BP_DW <double>] [-ANN_BP_MOMENT <double>]
  -FEATURES:<str>          	Features
	Grid list, input
  -NORMALIZE:<str>         	Normalize
	Boolean
	Default: 0
  -CLASSES:<str>           	Classification
	Grid, output
  -CLASSES_LUT:<str>       	Look-up Table
	Table, output, optional
  -MODEL_TRAIN:<str>       	Training
	Choice
	Available Choices:
	[0] training areas
	[1] training samples
	[2] load from file
	Default: 0
  -TRAIN_SAMPLES:<str>     	Training Samples
	Table, input
  -TRAIN_AREAS:<str>       	Training Areas
	Shapes, input
  -TRAIN_CLASS:<str>       	Class Identifier
	Table field
  -TRAIN_BUFFER:<double>   	Buffer Size
	Floating point
	Minimum: 0.000000
	Default: 1.000000
  -MODEL_LOAD:<str>        	Load Model
	File path
  -MODEL_SAVE:<str>        	Save Model
	File path
  -ANN_LAYERS:<num>        	Number of Layers
	Integer
	Minimum: 1
	Default: 3
  -ANN_NEURONS:<num>       	Number of Neurons
	Integer
	Minimum: 1
	Default: 5
  -ANN_MAXITER:<num>       	Maximum Number of Iterations
	Integer
	Minimum: 1
	Default: 300
  -ANN_EPSILON:<double>    	Error Change (Epsilon)
	Floating point
	Minimum: 0.000000
	Default: 0.000000
  -ANN_ACTIVATION:<str>    	Activation Function
	Choice
	Available Choices:
	[0] Identity
	[1] Sigmoid
	[2] Gaussian
	Default: 1
  -ANN_ACT_ALPHA:<double>  	Function's Alpha
	Floating point
	Default: 1.000000
  -ANN_ACT_BETA:<double>   	Function's Beta
	Floating point
	Default: 1.000000
  -ANN_PROPAGATION:<str>   	Training Method
	Choice
	Available Choices:
	[0] resilient propagation
	[1] back propagation
	Default: 1
  -ANN_RP_DW0:<double>     	Initial Update Value
	Floating point
	Default: 0.000000
  -ANN_RP_DW_PLUS:<double> 	Increase Factor
	Floating point
	Minimum: 1.010000
	Default: 1.200000
  -ANN_RP_DW_MINUS:<double>	Decrease Factor
	Floating point
	Minimum: 0.010000
	Maximum: 0.990000
	Default: 0.500000
  -ANN_RP_DW_MIN:<double>  	Lower Value Update Limit
	Floating point
	Minimum: 0.010000
	Default: 0.100000
  -ANN_RP_DW_MAX:<double>  	Upper Value Update Limit
	Floating point
	Minimum: 1.010000
	Default: 1.100000
  -ANN_BP_DW:<double>      	Weight Gradient Term
	Floating point
	Default: 0.100000
  -ANN_BP_MOMENT:<double>  	Moment Term
	Floating point
	Default: 0.100000