/// <summary>
/// Get an image reader to sequentially read images from disk for training.
/// </summary>
/// <param name="mapFilePath">The path to the map file</param>
/// <param name="imageWidth">The width to scale all images to</param>
/// <param name="imageHeight">The height to scale all images to</param>
/// <param name="numChannels">The number of channels to transform all images to</param>
/// <param name="numClasses">The number of label classes in this training set</param>
/// <param name="randomizeData">Set to true to randomize the data for training</param>
/// <param name="augmentData">Set to true to use data augmentation to expand the training set</param>
/// <returns>An image source ready for use in training or testing.</returns>
public static CNTK.MinibatchSource GetImageReader(string mapFilePath, int imageWidth, int imageHeight, int numChannels, int numClasses, bool randomizeData, bool augmentData)
{
var transforms = new List <CNTK.CNTKDictionary>();
if (augmentData)
{
var randomSideTransform = CNTK.CNTKLib.ReaderCrop("RandomSide",
new Tuple <int, int>(0, 0),
new Tuple <float, float>(0.8f, 1.0f),
new Tuple <float, float>(0.0f, 0.0f),
new Tuple <float, float>(1.0f, 1.0f),
"uniRatio");
transforms.Add(randomSideTransform);
}
var scaleTransform = CNTK.CNTKLib.ReaderScale(imageWidth, imageHeight, numChannels);
transforms.Add(scaleTransform);
var imageDeserializer = CNTK.CNTKLib.ImageDeserializer(mapFilePath, "labels", (uint)numClasses, "features", transforms);
var minibatchSourceConfig = new CNTK.MinibatchSourceConfig(new CNTK.DictionaryVector()
{
imageDeserializer
});
if (!randomizeData)
{
minibatchSourceConfig.randomizationWindowInChunks = 0;
minibatchSourceConfig.randomizationWindowInSamples = 0;
}
return(CNTK.CNTKLib.CreateCompositeMinibatchSource(minibatchSourceConfig));
}
CNTK.MinibatchSource create_minibatch_source(CNTK.NDShape shape, int start_index, int sample_count, string prefix, bool is_training = false, bool use_augmentations = false)
{
var map_filename = create_map_txt_file_if_needed(start_index, sample_count, prefix);
var transforms = new List <CNTK.CNTKDictionary>();
if (use_augmentations)
{
var randomSideTransform = CNTK.CNTKLib.ReaderCrop("RandomSide",
new Tuple <int, int>(0, 0),
new Tuple <float, float>(0.8f, 1.0f),
new Tuple <float, float>(0.0f, 0.0f),
new Tuple <float, float>(1.0f, 1.0f),
"uniRatio");
transforms.Add(randomSideTransform);
}
var scaleTransform = CNTK.CNTKLib.ReaderScale(width: shape[1], height: shape[0], channels: shape[2]);
transforms.Add(scaleTransform);
var imageDeserializer = CNTK.CNTKLib.ImageDeserializer(map_filename, "labels", 2, "features", transforms);
var minibatchSourceConfig = new CNTK.MinibatchSourceConfig(new CNTK.DictionaryVector()
{
imageDeserializer
});
if (!is_training)
{
minibatchSourceConfig.randomizationWindowInChunks = 0;
minibatchSourceConfig.randomizationWindowInSamples = 0;
}
return(CNTK.CNTKLib.CreateCompositeMinibatchSource(minibatchSourceConfig));
}
CNTK.MinibatchSource create_minibatchSource(string map_file, int num_classes, bool train)
{
var transforms = new List <CNTK.CNTKDictionary>();
if (true)
{
var randomSideTransform = CNTK.CNTKLib.ReaderCrop("RandomSide",
new Tuple <int, int>(0, 0),
new Tuple <float, float>(0.8f, 1.0f),
new Tuple <float, float>(0.0f, 0.0f),
new Tuple <float, float>(1.0f, 1.0f),
"uniRatio");
transforms.Add(randomSideTransform);
}
var scaleTransform = CNTK.CNTKLib.ReaderScale(image_width, image_height, num_channels);
transforms.Add(scaleTransform);
var imageDeserializer = CNTK.CNTKLib.ImageDeserializer(map_file, "labels", (uint)num_classes, "features", transforms);
var minibatchSourceConfig = new CNTK.MinibatchSourceConfig(new CNTK.DictionaryVector()
{
imageDeserializer
});
return(CNTK.CNTKLib.CreateCompositeMinibatchSource(minibatchSourceConfig));
}
