// Performs a prediction on a single image and loaded neural model; returns a predicted index
// For unknown reasons these single instance invoked Python predictions only like single threading
public int Predict(List <byte[]> image, string guid, bool isDebug)
{
if (model_loaded == null || dataSets == null)
{
throw new Exception("Neural Network not initialized!");
}
NDarray x_data = dataSets.GetDataSet(image);
if (config.isCNN)
{
x_data = (K.ImageDataFormat() == "channels_first")
? x_data.reshape(x_data.shape[0], 1, height, width)
: x_data.reshape(x_data.shape[0], height, width, 1);
}
NDarray y = model_loaded.Predict(x_data);
y = y.argmax();
int index = y.asscalar <int>();
bool result = dataSets.GetImageDatas().isLabelValid(index, guid, isDebug);
predicted += (Convert.ToInt32(result));
predictions++;
return(index);
}
static void Main(string[] args)
{
//Disable Eager Execution for tensorflow 1.15
K.DisableEager();
//ImageRecognitionApplication.Run();
MNIST_CNN.Run();
Cifar10_CNN.Run();
var cifar_pred = Cifar10_CNN.Predict("..\\..\\..\\img\\cifar_test2.jpg");
Console.WriteLine("cifar_test2.jpg is " + cifar_pred);
Console.ReadLine();
}
// Constructor
public NeuralNet(int width, int height, string log_dir, string log_ext, string model_dir, string model_ext, string config_ext)
{
this.width = width;
this.height = height;
this.log_dir = log_dir;
this.log_ext = log_ext;
this.model_dir = model_dir;
this.model_ext = model_ext;
this.config_ext = config_ext;
// Attempt to reset the invoked library
//np.arange(1);
//PythonEngine.BeginAllowThreads();
K.DisableEager();
K.ClearSession();
K.ResetUids();
}
// Performs predictions on a supplied dataset and loaded neural model:
// Calculates prediction accuracies and errors relative to the expected labeling
public ImageDatas Predict(bool isCNN)
{
if (model_loaded == null)
{
throw new Exception("Invalid Model!");
}
if (dataSets == null || !dataSets.isImageDatas())
{
throw new Exception("Invalid Dataset!");
}
ImageDatas ids = dataSets.GetImageDatas();
ImageDatas idf = new ImageDatas();
NDarray x_data = dataSets.BuildDataSet();
List <string> labels = ids.GetLabels();
if (isCNN)
{
x_data = (K.ImageDataFormat() == "channels_first")
? x_data.reshape(x_data.shape[0], 1, height, width)
: x_data.reshape(x_data.shape[0], height, width, 1);
}
Console.WriteLine("Predicting {0} Images", ids.Count);
NDarray y = model_loaded.Predict(x_data, verbose: 2);
int index;
NDarray result;
for (int i = 0; i < y.len; i++)
{
result = y[i];
result = result.argmax();
index = result.asscalar <int>();
if (ids[i].Label != labels[index])
{
ids[i].Index = labels.IndexOf(ids[i].Label) + 1;
idf.Add(ids[i]);
}
}
double accuracy = Math.Round(((y.len - idf.Count) * 100) / (double)y.len, 2);
idf.SetResults(string.Format("Predicted:{0} Correct: {1} Incorrect:{2} Accuracy:{3}", y.len, y.len - idf.Count, idf.Count, accuracy));
return(idf);
}
private Shape GetShape()
{
Shape shape;
if (K.ImageDataFormat() == "channels_first")
{
x_train = x_train.reshape(x_train.shape[0], 1, height, width);
x_test = x_test.reshape(x_test.shape[0], 1, height, width);
shape = (1, height, width);
}
else
{
x_train = x_train.reshape(x_train.shape[0], height, width, 1);
x_test = x_test.reshape(x_test.shape[0], height, width, 1);
shape = (height, width, 1);
}
return(shape);
}
// Neural model loading and saving routines
public void LoadModel(string stime)
{
K.DisableEager();
K.ClearSession();
K.ResetUids();
string filename = Utils.GetFileWithoutExtension(model_dir, stime);
config = (Config)FileIO.DeSerializeXml(typeof(Config), filename + config_ext);
if (config == null || string.IsNullOrEmpty(config.Model))
{
throw new Exception("Invalid config file");
}
model_loaded = BaseModel.ModelFromJson(config.Model);
model_loaded.LoadWeight(filename + model_ext);
model_loaded.Summary();
starttime = config.StartTime;
}
public static void Run()
{
int batch_size = 128;
int num_classes = 10;
int epochs = 12;
// input image dimensions
int img_rows = 28, img_cols = 28;
Shape input_shape = null;
// the data, split between train and test sets
var((x_train, y_train), (x_test, y_test)) = MNIST.LoadData();
if (K.ImageDataFormat() == "channels_first")
{
x_train = x_train.reshape(x_train.shape[0], 1, img_rows, img_cols);
x_test = x_test.reshape(x_test.shape[0], 1, img_rows, img_cols);
input_shape = (1, img_rows, img_cols);
}
else
{
x_train = x_train.reshape(x_train.shape[0], img_rows, img_cols, 1);
x_test = x_test.reshape(x_test.shape[0], img_rows, img_cols, 1);
input_shape = (img_rows, img_cols, 1);
}
x_train = x_train.astype(np.float32);
x_test = x_test.astype(np.float32);
x_train /= 255;
x_test /= 255;
Console.WriteLine("x_train shape: " + x_train.shape);
Console.WriteLine(x_train.shape[0] + " train samples");
Console.WriteLine(x_test.shape[0] + " test samples");
// convert class vectors to binary class matrices
y_train = Util.ToCategorical(y_train, num_classes);
y_test = Util.ToCategorical(y_test, num_classes);
// Build CNN model
var model = new Sequential();
model.Add(new Conv2D(32, kernel_size: (3, 3).ToTuple(),
activation: "relu",
input_shape: input_shape));
model.Add(new Conv2D(64, (3, 3).ToTuple(), activation: "relu"));
model.Add(new MaxPooling2D(pool_size: (2, 2).ToTuple()));
model.Add(new Dropout(0.25));
model.Add(new Flatten());
model.Add(new Dense(128, activation: "relu"));
model.Add(new Dropout(0.5));
model.Add(new Dense(num_classes, activation: "softmax"));
model.Compile(loss: "categorical_crossentropy",
optimizer: new Adadelta(), metrics: new string[] { "accuracy" });
model.Fit(x_train, y_train,
batch_size: batch_size,
epochs: epochs,
verbose: 1,
validation_data: new NDarray[] { x_test, y_test });
var score = model.Evaluate(x_test, y_test, verbose: 0);
Console.WriteLine("Test loss:" + score[0]);
Console.WriteLine("Test accuracy:" + score[1]);
}
static void Main(string[] args)
{
int batch_size = 128; //Training batch size
int num_classes = 10; //No. of classes
int epochs = 12; //No. of epoches we will train
// input image dimensions
int img_rows = 28, img_cols = 28;
// Declare the input shape for the network
Shape input_shape = null;
// Load the MNIST dataset into Numpy array
var((x_train, y_train), (x_test, y_test)) = MNIST.LoadData();
//Check if its channel fist or last and rearrange the dataset accordingly
if (K.ImageDataFormat() == "channels_first")
{
x_train = x_train.reshape(x_train.shape[0], 1, img_rows, img_cols);
x_test = x_test.reshape(x_test.shape[0], 1, img_rows, img_cols);
input_shape = (1, img_rows, img_cols);
}
else
{
x_train = x_train.reshape(x_train.shape[0], img_rows, img_cols, 1);
x_test = x_test.reshape(x_test.shape[0], img_rows, img_cols, 1);
input_shape = (img_rows, img_cols, 1);
}
//Normalize the input data
x_train = x_train.astype(np.float32);
x_test = x_test.astype(np.float32);
x_train /= 255;
x_test /= 255;
Console.WriteLine("x_train shape: " + x_train.shape);
Console.WriteLine(x_train.shape[0] + " train samples");
Console.WriteLine(x_test.shape[0] + " test samples");
// Convert class vectors to binary class matrices
y_train = Util.ToCategorical(y_train, num_classes);
y_test = Util.ToCategorical(y_test, num_classes);
// Build CNN model
var model = new Sequential();
model.Add(new Conv2D(32, kernel_size: (3, 3).ToTuple(),
activation: "relu",
input_shape: input_shape));
model.Add(new Conv2D(64, (3, 3).ToTuple(), activation: "relu"));
model.Add(new MaxPooling2D(pool_size: (2, 2).ToTuple()));
model.Add(new Dropout(0.25));
model.Add(new Flatten());
model.Add(new Dense(128, activation: "relu"));
model.Add(new Dropout(0.5));
model.Add(new Dense(num_classes, activation: "softmax"));
//Compile with loss, metrics and optimizer
model.Compile(loss: "categorical_crossentropy",
optimizer: new Adadelta(), metrics: new string[] { "accuracy" });
//Train the model
model.Fit(x_train, y_train,
batch_size: batch_size,
epochs: epochs,
verbose: 1,
validation_data: new NDarray[] { x_test, y_test });
//Score the model for performance
var score = model.Evaluate(x_test, y_test, verbose: 0);
Console.WriteLine("Test loss:" + score[0]);
Console.WriteLine("Test accuracy:" + score[1]);
// Save the model to HDF5 format which can be loaded later or ported to other application
model.Save("model.h5");
// Save it to Tensorflow JS format and we will test it in browser.
var v = K.Instance;
//model.SaveTensorflowJSFormat(@"C:\_temp\");
//model.SaveOnnx(@"C:\_temp\");
Console.ReadLine();
}