mibDeepSupport Directory Reference

Directory dependency graph for mibDeepSupport:

Files
	customDiceForwardLoss.m
	loss = customDiceForwardLoss(Y, T, dataDimension, useClasses) returns the Dice loss between the predictions Y and the training targets T.
	exportONNXNetwork.m
	exportONNXNetwork Export network or graph of layers to ONNX model format.
	matReadInstanceLabels.m
	read preprocessed instance segmentation labels and return them as a cell array with 3 element.
	mibDeepAugmentAndCrop2dPatchMultiGPU.m
	mibDeepAugmentAndCrop3dPatchMultiGPU.m
	mibDeepConvertOldAugmentationSettingsToNew.m
	supporting function for DeepMIB, the function converts old augmentation settings to the new set of settings from MIB 2.8455
	mibDeepCustomTrainingProgressDisplay.m
	mibDeepCustomTrainingProgressDisplayTrainNet.m
	mibDeepGenerateDefaultAugmentationSettings.m
	generate default augmentation settings from MIB 2.8455
	mibDeepSaveTrainingPlot.m
	save custom training plot to a file
	mibDeepStopTrainingCallback.m
	stop training process by pressing the Stop training or Emergency stop buttons
	mibDeepStopTrainingWithoutPlots.m
	stop training, when it is done without the training plot
	mibDeepStoreLoadCategorical.m
	supporting function for mibDeepController read categorical dataset and return it as a cell similar to pixelLabelDatastore
	mibDeepStoreLoadImages.m
	supporting function for mibDeepController load image function for the imagedatastore
	mibDeepStoreLoadModel.m
	supporting function for mibDeepController to be used with pixelLabelDatastore to load MIB models
	mibDeepTrainingProgressStructUpdateAxesLimits.m
	function to handle callbacks from the context menu for mibDeepTrainingProgressStruct.UILossAxes
	saveInstanceLabelsParFor.m
	saveInstanceLabelsParFor(fn, imageFilename, instanceBoxes, instanceNames, instanceMasks, compressModels) saving from parfor loop converted labels for training of 2D instance segmentation as it is not possible to use save directly inside parfor. This function is used in mibDeepController.processImagesForInstanceSegmentation to preprocess instance labels