/// <summary>
/// Constructor
/// </summary>
/// <param name="baseFolder">Base folder with MNIST dataset files</param>
/// <param name="imagesCnt">How many images to load</param>
/// <param name="exact">If TRUE, you will get exactly imagesCnt images. If FALSE, you will get AT MOST imagesCnt images. It's here for performance reason and BC</param>
/// <param name="afterLastImage">What to do, after the last values has been sent</param>
public MyMNISTManager(string baseFolder, int trainingExamplesPerDigitCnt = int.MaxValue, int testExamplesPerDigitCnt = int.MaxValue,
bool exact = false, MNISTLastImageMethod afterLastImage = MNISTLastImageMethod.ResetToStart)
{
m_baseFolder = baseFolder;
m_trainingImages = new ArrayList();
m_testImages = new ArrayList();
m_afterLastImage = afterLastImage;
m_trainingExamplesPerDigitCnt = trainingExamplesPerDigitCnt;
m_testExamplesPerDigitCnt = testExamplesPerDigitCnt;
m_imagesServed = 0;
m_sequenceIterator = 0;
m_definedOrder = false;
// START TIME MEASUREMENT ----------------------------------------------------------
//Stopwatch sw = new Stopwatch();
//sw.Start();
ReadMnistSet("train-images.idx3-ubyte", "train-labels.idx1-ubyte", m_trainingExamplesPerDigitCnt, m_trainingImages);
ReadMnistSet("t10k-images.idx3-ubyte", "t10k-labels.idx1-ubyte", m_testExamplesPerDigitCnt, m_testImages);
byte[,] data = new byte[28, 28];
m_blankImage = new MyMNISTImage(data, 0);
m_trainingImagesEnumerator = m_trainingImages.GetEnumerator();
m_testImagesEnumerator = m_testImages.GetEnumerator();
//sw.Stop();
//Console.WriteLine("Elapsed={0}", sw.Elapsed);
}
/// <summary>
/// Constructor
/// </summary>
/// <param name="baseFolder">Base folder with MNIST dataset files</param>
/// <param name="imagesCnt">How many images to load</param>
/// <param name="exact">If TRUE, you will get exactly imagesCnt images. If FALSE, you will get AT MOST imagesCnt images. It's here for performance reason and BC</param>
/// <param name="afterLastImage">What to do, after the last values has been sent</param>
public MyMNISTManager(string baseFolder, int trainingExamplesPerDigitCnt = int.MaxValue, int testExamplesPerDigitCnt = int.MaxValue,
bool exact = false, MNISTLastImageMethod afterLastImage = MNISTLastImageMethod.ResetToStart, int randomSeed = 0)
{
if (randomSeed == 0)
{
rand = new Random();
}
else
{
rand = new Random(randomSeed);
}
m_baseFolder = baseFolder;
m_trainingImages = new ArrayList();
m_testImages = new ArrayList();
m_afterLastImage = afterLastImage;
m_trainingExamplesPerDigitCnt = trainingExamplesPerDigitCnt;
m_testExamplesPerDigitCnt = testExamplesPerDigitCnt;
m_lastServedImage = -1;
m_sequenceIterator = 0;
m_definedOrder = false;
// START TIME MEASUREMENT ----------------------------------------------------------
//Stopwatch sw = new Stopwatch();
//sw.Start();
ReadMnistSet("train-images.idx3-ubyte", "train-labels.idx1-ubyte", m_trainingExamplesPerDigitCnt, m_trainingImages);
ReadMnistSet("t10k-images.idx3-ubyte", "t10k-labels.idx1-ubyte", m_testExamplesPerDigitCnt, m_testImages);
byte[,] data = new byte[28, 28];
m_blankImage = new MyMNISTImage(data, 0);
m_trainingImagesEnumerator = m_trainingImages.GetEnumerator();
m_testImagesEnumerator = m_testImages.GetEnumerator();
//sw.Stop();
//Console.WriteLine("Elapsed={0}", sw.Elapsed);
}
/// <summary>
/// Constructor
/// </summary>
/// <param name="baseFolder">Base folder with MNIST dataset files</param>
/// <param name="imagesCnt">How many images to load</param>
/// <param name="exact">If TRUE, you will get exactly imagesCnt images. If FALSE, you will get AT MOST imagesCnt images. It's here for performance reason and BC</param>
/// <param name="afterLastImage">What to do, after the last values has been sent</param>
public MyMNISTManager(string baseFolder, int imagesCnt = int.MaxValue, bool exact = false, MNISTLastImageMethod afterLastImage = MNISTLastImageMethod.ResetToStart)
{
m_baseFolder = baseFolder;
m_images = new ArrayList();
m_afterLastImage = afterLastImage;
m_imagesDemand = imagesCnt;
m_imagesServed = 0;
m_sequenceIterator = 0;
m_definedOrder = false;
FileStream ifsTrainLabels = new FileStream(m_baseFolder + "train-labels.idx1-ubyte", FileMode.Open, FileAccess.Read);
FileStream ifsTrainImages = new FileStream(m_baseFolder + "train-images.idx3-ubyte", FileMode.Open, FileAccess.Read);
BinaryReader brTrainLabels = new BinaryReader(ifsTrainLabels);
BinaryReader brTrainImages = new BinaryReader(ifsTrainImages);
//Magic number
brTrainLabels.ReadInt32();
brTrainImages.ReadInt32();
int numImagesTrainLables = brTrainLabels.ReadInt32();
numImagesTrainLables = ReverseBytes(numImagesTrainLables);
int numImagesTrainImages = brTrainImages.ReadInt32();
numImagesTrainImages = ReverseBytes(numImagesTrainImages);
int numRowsTrainImages = brTrainImages.ReadInt32();
numRowsTrainImages = ReverseBytes(numRowsTrainImages);
int numColsTrainImages = brTrainImages.ReadInt32();
numColsTrainImages = ReverseBytes(numColsTrainImages);
int maxImages;
if (exact)
{
// numImagesTrainImages = 60000
maxImages = numImagesTrainImages;
}
else
{
// value of 2000 is a compromise between long loading-time for maximum of 60k images and minimum of user set "imagesCnt"
// this also brings more flexibility for changing the m_imagesDemand during simulation (up to max(2000,imagesCnt) instead of imagesCnt)
maxImages = Math.Max(2000, Math.Min(numImagesTrainImages, imagesCnt));
}
byte[][] data = new byte[numColsTrainImages][];
for (int i = 0; i < data.Length; ++i)
{
data[i] = new byte[numRowsTrainImages];
}
for (int i = 0; i < maxImages; ++i)
{
for (int j = 0; j < numColsTrainImages; ++j)
{
for (int k = 0; k < numRowsTrainImages; ++k)
{
byte b = brTrainImages.ReadByte();
data[j][k] = b;
}
}
byte label = brTrainLabels.ReadByte();
MyMNISTImage mImage = new MyMNISTImage(data, label, numColsTrainImages, numRowsTrainImages);
m_images.Add(mImage);
}
ifsTrainImages.Close();
brTrainImages.Close();
ifsTrainLabels.Close();
brTrainLabels.Close();
m_imageEnumerator = m_images.GetEnumerator();
for (int i = 0; i < numColsTrainImages; ++i)
{
for (int j = 0; j < numRowsTrainImages; ++j)
{
data[i][j] = 0;
}
}
m_blankImage = new MyMNISTImage(data, 0);
}
