def train(holdout=False, holdout_list=[]):
# load data
X, Y = load_X_and_Y()
x_train, x_dev, x_test = X
y_train, y_dev, y_test = Y
# holdout training and dev data if requested
if holdout:
X_descr = load_X_descr()
x_train_all_descr, x_dev_all_descr, _ = X_descr
holdout_x_train = []
holdout_y_train = []
for idx in range(len(x_train)):
if x_train_all_descr[idx] not in holdout_list:
holdout_x_train.append(x_train[idx])
holdout_y_train.append(y_train[idx])
holdout_x_dev = []
holdout_y_dev = []
for idx in range(len(x_dev)):
if x_dev_all_descr[idx] not in holdout_list:
holdout_x_dev.append(x_dev[idx])
holdout_y_dev.append(y_dev[idx])
x_train = np.array(holdout_x_train).reshape((-1, 224, 224, 3))
y_train = np.array(holdout_y_train).reshape((-1, 1))
x_dev = np.array(holdout_x_dev).reshape((-1, 224, 224, 3))
y_dev = np.array(holdout_y_dev).reshape((-1, 1))
# train model
model = CNN()
model.fit(x_train, y_train, x_dev, y_dev, save=True)
model.evaluate(x_test, y_test)
def train(holdout=False, holdout_list=[]):
# load data
X, Y = load_X_and_Y()
x_train, x_dev, x_test = X
y_train, y_dev, y_test = Y
# holdout training and dev data if requested
if holdout:
X_descr = load_X_descr()
x_train_all_descr, x_dev_all_descr, _ = X_descr
holdout_x_train = []
holdout_y_train = []
for idx in range(len(x_train)):
if x_train_all_descr[idx] not in holdout_list:
holdout_x_train.append(x_train[idx])
holdout_y_train.append(y_train[idx])
holdout_x_dev = []
holdout_y_dev = []
for idx in range(len(x_dev)):
if x_dev_all_descr[idx] not in holdout_list:
holdout_x_dev.append(x_dev[idx])
holdout_y_dev.append(y_dev[idx])
x_train = np.array(holdout_x_train).reshape((-1, 224, 224, 3))
y_train = np.array(holdout_y_train).reshape((-1, 1))
x_dev = np.array(holdout_x_dev).reshape((-1, 224, 224, 3))
y_dev = np.array(holdout_y_dev).reshape((-1, 1))
# train model
model = CNN()
model.fit(x_train, y_train, x_dev, y_dev, save=True)
model.evaluate(x_test, y_test)
def train():
# load data
X, Y = load_X_and_Y()
x_train, x_dev, x_test = X
y_train, y_dev, y_test = Y
# train model
model = CNN()
model.fit(x_train, y_train, x_dev, y_dev, save=True)
model.evaluate(x_test, y_test)
def test_train():
from tensorflow.keras.datasets import mnist
import tensorflow.keras as keras
bsize_entity = 128
no_of_cls = 10
epochs = 1
img_row, img_col = 28, 28
(x_train, y_train), (x_test, y_test) = mnist.load_data()
x_train = x_train.reshape(60000, 28, 28, 1)
x_test = x_test.reshape(10000, 28, 28, 1)
y_train = keras.utils.to_categorical(y_train, no_of_cls)
y_test = keras.utils.to_categorical(y_test, no_of_cls)
model = CNN()
model.add_input_layer(shape=(28, 28, 1), name="Input")
model.append_conv2d_layer(32, kernel_size=(3, 3))
model.append_conv2d_layer(64, kernel_size=(3, 3))
model.append_maxpooling2d_layer(pool_size=(2, 2))
model.append_flatten_layer()
model.append_dense_layer(128, activation="relu")
model.append_dense_layer(no_of_cls, activation="softmax")
model.set_loss_function("categorical_crossentropy")
model.set_metric("accuracy")
model.set_optimizer("Adagrad")
model.train(x_train, y_train, batch_size=bsize_entity, num_epochs=epochs)
mk = model.evaluate(x_test, y_test)
correct = np.array([0.06997422293154523, 0.9907])
np.testing.assert_almost_equal(correct[0], mk[0], decimal=2)
np.testing.assert_almost_equal(correct[1], mk[1], decimal=2)
def test(self):
'''
use self.clf to get score/accuracy
prints accuracy
draws plot
'''
img_rows,img_columns = 45,45
test_data = self.test_img.reshape((self.test_img.shape[0], img_rows,img_columns))
test_data = test_data[:, np.newaxis, :, :]
label_map={}
count = 0
for folder in os.listdir("./data"):
label_map[folder]=count
count+=1
for j in range(len(self.test_labels)):
self.test_labels[j]=label_map[self.test_labels[j]]
total_classes = count
test_labels = np_utils.to_categorical(self.test_labels,total_classes)
sgd = optimizers.SGD(lr=0.01, decay=1e-6, momentum=0.9, nesterov=True)
clf = CNN().build(img_rows,img_columns,1,total_classes,'model.h5')
clf.compile(loss="categorical_crossentropy", optimizer=sgd, metrics=["accuracy"])
loss, accuracy = clf.evaluate(test_data, test_labels, batch_size=self.b_size, verbose=1)
print('Accuracy of Model: {:.2f}%'.format(accuracy * 100))
return accuracy
def test_evaluate():
from tensorflow.keras.datasets import cifar10
(X_train, y_train), (X_test, y_test) = cifar10.load_data()
X_train = X_train[:100, :]
y_train = y_train[:100]
X_test=X_test[:100,:]
y_test = y_test[:100]
# np.random.seed(100)
initilizer = tensorflow.keras.initializers.Zeros()
# initilizer = tensorflow.keras.initializers.RandomUniform(minval=-0.05, maxval=0.05, seed=20)
model_testing = CNN()
model = Sequential()
model.add(Conv2D(filters=64, kernel_size=3, strides=1, padding='same', activation='relu', trainable=True,
input_shape=(32, 32, 3), kernel_initializer=initilizer, bias_initializer=initilizer))
model.add(Conv2D(filters=70, kernel_size=3, strides=1, padding='same', activation='relu', trainable=True,
kernel_initializer=initilizer, bias_initializer=initilizer))
model.add(Conv2D(filters=75, kernel_size=3, strides=1, padding='same', activation='relu', trainable=True,
kernel_initializer=initilizer, bias_initializer=initilizer))
model.add(Conv2D(filters=90, kernel_size=3, strides=1, padding='same', activation='relu', trainable=True,
kernel_initializer=initilizer, bias_initializer=initilizer))
model.add(MaxPool2D(pool_size=2, padding='same', strides=1))
model.add(Flatten())
model.add(
Dense(units=256, activation='relu', trainable=True, kernel_initializer=initilizer, bias_initializer=initilizer))
model.add(
Dense(units=256, activation='relu', trainable=True, kernel_initializer=initilizer, bias_initializer=initilizer))
model.add(Dense(units=256, activation='sigmoid', trainable=True, kernel_initializer=initilizer,
bias_initializer=initilizer))
model.compile(optimizer='Adagrad', loss='hinge', metrics=['mse'])
history = model.fit(x=X_train, y=y_train, batch_size=32, epochs=5, shuffle=False)
model_testing.add_input_layer(shape=(32, 32, 3), name="")
model_testing.append_conv2d_layer(num_of_filters=64, kernel_size=3, padding='same', strides=1, activation='relu',
name="1")
model_testing.append_conv2d_layer(num_of_filters=70, kernel_size=3, padding='same', strides=1, activation='relu',
name="2")
model_testing.append_conv2d_layer(num_of_filters=75, kernel_size=3, padding='same', strides=1, activation='relu',
name="3")
model_testing.append_conv2d_layer(num_of_filters=90, kernel_size=3, padding='same', strides=1, activation='relu',
name="4")
model_testing.append_maxpooling2d_layer(pool_size=2, padding='same', strides=1, name="5")
model_testing.append_flatten_layer(name='6')
model_testing.append_dense_layer(num_nodes=256, activation='relu', name='7')
model_testing.append_dense_layer(num_nodes=256, activation='relu', name='8')
model_testing.append_dense_layer(num_nodes=256, activation='sigmoid', name='9')
model_testing.set_optimizer(optimizer='adagrad')
model_testing.set_loss_function(loss='hinge')
model_testing.set_metric(metric='mse')
loss = model_testing.train(X_train=X_train, y_train=y_train, batch_size=32, num_epochs=5)
model_evaluate = model.evaluate(X_test,y_test)
model_testing_evaluate = model_testing.evaluate(X_test,y_test)
#assert model_testing_evaluate == model_evaluate
assert np.allclose(model_testing_evaluate,model_evaluate,rtol=1e-2,atol=1e-2)
def test_evaluate():
cnn = CNN()
batch_size = 10
num_epochs = 10
(X_train, y_train), (X_test, y_test) = cifar10.load_data()
samples = 100
X_train = X_train[0:samples, :]
X_train = X_train.astype('float32') / 255
test_samples = 10
X_test = X_test[0:test_samples, :]
X_test = X_test.astype('float32') / 255
from tensorflow.keras.utils import to_categorical
y_train = to_categorical(y_train, 10)
y_train = y_train[0:samples, :]
y_test = to_categorical(y_test, 10)
y_test = y_test[0:test_samples, :]
cnn.add_input_layer(shape=(32, 32, 3))
cnn.append_conv2d_layer(num_of_filters=64, kernel_size=(3, 3), activation='relu', name="conv1")
cnn.append_conv2d_layer(num_of_filters=32, kernel_size=(3, 3), activation='relu', name="conv2")
cnn.append_flatten_layer(name="flat1")
cnn.append_dense_layer(num_nodes=10, activation="relu", name="dense1")
cnn.set_optimizer(optimizer="SGD")
cnn.set_loss_function(loss="hinge")
cnn.set_metric(metric='accuracy')
ListofLoss = cnn.train(X_train=X_train, y_train=y_train, batch_size=None, num_epochs=num_epochs)
(loss, metric) = cnn.evaluate(X=X_test, y=y_test)
assert loss < 5
assert metric < 2
def test_evaluate():
(X_train, y_train), (X_test, y_test) = cifar10.load_data()
number_of_train_samples_to_use = 500
number_of_test_samples_to_use = 200
X_train = X_train[0:number_of_train_samples_to_use, :]
y_train = y_train[0:number_of_train_samples_to_use]
X_test = X_test[0:number_of_test_samples_to_use, :]
y_test = y_test[0:number_of_test_samples_to_use]
my_cnn = CNN()
my_cnn.add_input_layer(shape=(32, 32, 3), name="input")
my_cnn.append_conv2d_layer(num_of_filters=32,
kernel_size=(3, 3),
padding="same",
activation='relu',
name="conv1")
my_cnn.append_maxpooling2d_layer(pool_size=(2, 2), name="pool1")
my_cnn.append_conv2d_layer(num_of_filters=64,
kernel_size=3,
activation='relu',
name="conv2")
my_cnn.append_maxpooling2d_layer(pool_size=(2, 2), name="pool2")
my_cnn.append_conv2d_layer(num_of_filters=64,
kernel_size=3,
activation='relu',
name="conv3")
my_cnn.append_flatten_layer(name="flat1")
my_cnn.append_dense_layer(num_nodes=64, activation="relu", name="dense1")
my_cnn.append_dense_layer(num_nodes=10,
activation="softmax",
name="dense2")
np.random.seed(seed=1)
weigh = my_cnn.get_weights_without_biases(layer_name="conv1")
w_set = np.random.rand(*weigh.shape)
my_cnn.set_weights_without_biases(w_set, layer_name="conv1")
np.random.seed(seed=1)
weigh = my_cnn.get_weights_without_biases(layer_name="conv2")
w_set = np.random.rand(*weigh.shape)
my_cnn.set_weights_without_biases(w_set, layer_name="conv2")
np.random.seed(seed=1)
weigh = my_cnn.get_weights_without_biases(layer_name="conv3")
w_set = np.random.rand(*weigh.shape)
my_cnn.set_weights_without_biases(w_set, layer_name="conv3")
np.random.seed(seed=1)
weigh = my_cnn.get_weights_without_biases(layer_name="dense1")
w_set = np.random.rand(*weigh.shape)
my_cnn.set_weights_without_biases(w_set, layer_name="dense1")
np.random.seed(seed=1)
weigh = my_cnn.get_weights_without_biases(layer_name="dense2")
w_set = np.random.rand(*weigh.shape)
my_cnn.set_weights_without_biases(w_set, layer_name="dense2")
my_cnn.set_loss_function()
my_cnn.set_optimizer(optimizer="SGD", learning_rate=0.01, momentum=0.0)
my_cnn.set_metric(metric="accuracy")
# los = np.array([2.30277, 2.30264, 2.30242, 2.30225, 2.30207, 2.30190, 2.30171, 2.30154, 2.30138])
# los = np.around(los,decimals=2)
my_cnn.train(X_train, y_train, 60, 10)
acc = my_cnn.evaluate(X_test, y_test)
de = np.float32(0.07)
assert (acc == de)
def test_train_and_evaluate():
from tensorflow.keras.datasets import cifar10
batch_size = 32
num_classes = 10
epochs = 1
save_dir = os.path.join(os.getcwd(), 'saved_models')
model_name = 'keras_cifar10_trained_model.h5'
(train_images, train_labels), (test_images,
test_labels) = cifar10.load_data()
train_labels = keras.utils.to_categorical(train_labels, num_classes)
test_labels = keras.utils.to_categorical(test_labels, num_classes)
my_cnn = CNN()
my_cnn.append_conv2d_layer(num_of_filters=32,
kernel_size=(3, 3),
padding="same",
activation='relu',
name="conv1",
input_shape=train_images.shape[1:])
# my_cnn.append_conv2d_layer(num_of_filters=32, kernel_size=(3,3),padding="same", activation='relu', name="conv2")
my_cnn.append_maxpooling2d_layer(pool_size=2,
padding="same",
strides=2,
name="pool1")
# my_cnn.append_conv2d_layer(num_of_filters=64, kernel_size=(3,3),padding="same", activation='relu', name="conv3")
my_cnn.append_conv2d_layer(num_of_filters=64,
kernel_size=(3, 3),
padding="same",
activation='relu',
name="conv4")
my_cnn.append_maxpooling2d_layer(pool_size=2,
padding="same",
strides=2,
name="pool2")
my_cnn.append_flatten_layer(name="flat1")
# my_cnn.append_dense_layer(num_nodes=512,activation="relu",name="dense1")
my_cnn.append_dense_layer(num_nodes=10,
activation="softmax",
name="dense2")
my_cnn.set_metric('accuracy')
my_cnn.set_optimizer('RMSprop')
my_cnn.set_loss_function('categorical_crossentropy')
loss = my_cnn.train(train_images, train_labels, batch_size, epochs)
path = os.getcwd()
file_path = os.path.join(path, model_name)
my_cnn.save_model(model_file_name=file_path)
print("loss :{0}".format(loss))
assert len(loss) == 1
test_loss, test_acc = my_cnn.evaluate(test_images, test_labels)
assert test_loss < 5
assert test_acc < 1
def test_evaluate():
from tensorflow.keras.datasets import mnist
import tensorflow.keras as keras
batch_size = 128
num_classes = 10
epochs = 3
# input image dimensions
img_rows, img_cols = 28, 28
# the data, split between train and test sets
(x_train, y_train), (x_test, y_test) = mnist.load_data()
x_train = x_train.reshape(60000, 28, 28, 1)
x_test = x_test.reshape(10000, 28, 28, 1)
print('x_train shape:', x_train.shape)
print(x_train.shape[0], 'train samples')
print(x_test.shape[0], 'test samples')
# convert class vectors to binary class matrices
y_train = keras.utils.to_categorical(y_train, num_classes)
y_test = keras.utils.to_categorical(y_test, num_classes)
model = CNN()
model.add_input_layer(shape=(28, 28, 1), name="Input")
model.append_conv2d_layer(32, kernel_size=(3, 3))
model.append_conv2d_layer(64, kernel_size=(3, 3))
model.append_maxpooling2d_layer(pool_size=(2, 2))
model.append_flatten_layer()
model.append_dense_layer(128, activation="relu")
model.append_dense_layer(num_classes, activation="softmax")
model.set_loss_function("categorical_crossentropy")
model.set_metric("accuracy")
model.set_optimizer("Adagrad")
model.train(x_train, y_train, batch_size=batch_size, num_epochs=epochs)
score = model.evaluate(x_test, y_test)
actual = np.array([0.05093684684933396, 0.9907])
# loss
np.testing.assert_almost_equal(actual[0], score[0], decimal=2)
# accuracy
np.testing.assert_almost_equal(actual[1], score[1], decimal=2)
def gridSearch(xTrain, yTrain, xDev, yDev, options):
paramCombos = myProduct(options)
bestCombo, bestCrossEntropy = None, float('inf')
scores = {}
for combo in paramCombos:
cnn = CNN(numFilters=combo['numFilters'], windowSize=combo['windowSize'])
cnn.fit(xTrain[:combo['numTrain']], yTrain[:combo['numTrain']], numEpochs=combo['numEpochs'],
batchSize=combo['batchSize'], verbose=True)
crossEntropy, accuracy = cnn.evaluate(xDev, yDev, showAccuracy=True)
scores[tuple(combo.items())] = (crossEntropy, accuracy)
if crossEntropy < bestCrossEntropy:
bestCombo, bestCrossEntropy = combo, crossEntropy
print 'Combo: {}, CE: {}, accuracy: {}'.format(combo, crossEntropy, accuracy)
return scores
def train(self):
#Change data to required format
img_rows,img_columns = 45,45
train_data = self.train_img.reshape((self.train_img.shape[0], img_rows,img_columns))
train_data = train_data[:, np.newaxis, :, :]
test_data = self.test_img.reshape((self.test_img.shape[0], img_rows,img_columns))
test_data = test_data[:, np.newaxis, :, :]
label_map={}
count = 0
for folder in os.listdir("../src/data"):
label_map[folder]=count
count+=1
for i in range(len(self.train_labels)):
self.train_labels[i]=label_map[self.train_labels[i]]
for j in range(len(self.test_labels)):
self.test_labels[j]=label_map[self.test_labels[j]]
# Transform training and testing data to 10 classes in range [0,classes] ; num. of classes = 0 to 9 = 10 classes
total_classes = count
train_labels = np_utils.to_categorical(self.train_labels, total_classes)
test_labels = np_utils.to_categorical(self.test_labels,total_classes)
# Defing and compile the SGD optimizer and CNN model
print('\n Compiling model...')
sgd = optimizers.SGD(lr=0.01, decay=1e-6, momentum=0.9, nesterov=True)
clf = CNN().build(img_rows,img_columns,1,total_classes)
clf.compile(loss="categorical_crossentropy", optimizer=sgd, metrics=["accuracy"])
# Initially train and test the model; If weight saved already, load the weights using arguments.
num_epoch = self.num_epoch # Number of epochs
verb = 1 # Verbose
print('\nTraining the Model...')
model_info=clf.fit(train_data, train_labels, batch_size=self.b_size, nb_epoch=num_epoch,verbose=verb)
# Evaluate accuracy and loss function of test data
print('Evaluating Accuracy and Loss Function...')
loss, accuracy = clf.evaluate(test_data, test_labels, batch_size=self.b_size, verbose=1)
print('Accuracy of Model',(accuracy * 100))
clf.save_weights('model.h5', overwrite=True)
print(model_info.history)
self.plot_model_history(model_info)
return accuracy
def gridSearch(xTrain, yTrain, xDev, yDev, options):
paramCombos = myProduct(options)
bestCombo, bestCrossEntropy = None, float('inf')
scores = {}
for combo in paramCombos:
cnn = CNN(numFilters=combo['numFilters'],
windowSize=combo['windowSize'])
cnn.fit(xTrain[:combo['numTrain']],
yTrain[:combo['numTrain']],
numEpochs=combo['numEpochs'],
batchSize=combo['batchSize'],
verbose=True)
crossEntropy, accuracy = cnn.evaluate(xDev, yDev, showAccuracy=True)
scores[tuple(combo.items())] = (crossEntropy, accuracy)
if crossEntropy < bestCrossEntropy:
bestCombo, bestCrossEntropy = combo, crossEntropy
print 'Combo: {}, CE: {}, accuracy: {}'.format(combo, crossEntropy,
accuracy)
return scores
def test_train_and_evaluate():
# Initializing and adding layers
print("*********** PLEASE WAIT FOR DATA TO LOAD ***********")
(train_images, train_labels), (test_images,
test_labels) = cifar10.load_data()
train_images, test_images = train_images / 255.0, test_images / 255.0
new_cnn = CNN()
new_cnn.add_input_layer(shape=(32, 32, 3), name="input")
new_cnn.append_conv2d_layer(32,
strides=3,
activation="relu",
name="conv2d_1")
new_cnn.append_maxpooling2d_layer(pool_size=2, name="maxpool_1")
new_cnn.append_conv2d_layer(64,
strides=3,
activation="relu",
name="conv2d_2")
new_cnn.append_maxpooling2d_layer(pool_size=2, name="maxpool_2")
new_cnn.append_conv2d_layer(64,
strides=3,
activation="relu",
name="conv2d_3")
new_cnn.append_flatten_layer(name="flatten")
new_cnn.append_dense_layer(64, activation="relu", name="dense_1")
new_cnn.append_dense_layer(10, activation="softmax", name="dense_2")
# Setting Compiler values
new_cnn.set_loss_function(loss="SparseCategoricalCrossentropy")
new_cnn.set_optimizer(optimizer="SGD")
new_cnn.set_metric('accuracy')
# Entering Num Epoch
batch_size = 1000
num_epoch = 10
history = new_cnn.train(train_images,
train_labels,
batch_size=batch_size,
num_epochs=num_epoch)
assert len(history) == len(train_images) / batch_size * num_epoch
evaluate = new_cnn.evaluate(train_images, train_labels)
assert evaluate[1] <= 1
assert evaluate[0] <= 3
model.append_conv2d_layer(num_of_filters=32,
kernel_size=3,
padding='same',
activation='relu')
model.append_conv2d_layer(num_of_filters=32,
kernel_size=3,
padding='same',
activation='relu')
model.append_maxpooling2d_layer(pool_size=2)
model.append_conv2d_layer(num_of_filters=64,
kernel_size=3,
padding='same',
activation='relu')
model.append_conv2d_layer(num_of_filters=64,
kernel_size=3,
padding='same',
activation='relu')
model.append_maxpooling2d_layer(pool_size=2)
model.append_flatten_layer()
model.append_dense_layer(num_nodes=512, activation='relu')
model.append_dense_layer(num_nodes=num_classes, activation='softmax')
model.set_optimizer(optimizer='RMSprop', learning_rate=0.0001)
model.set_loss_function('categorical_crossentropy')
model.set_metric(['accuracy'])
x_train = x_train.astype('float32')
x_test = x_test.astype('float32')
x_train /= 255
x_test /= 255
model.train(x_train, y_train, num_classes, epochs)
model.evaluate(x_test, y_test)
import numpy as np
convNet = CNN("AlphaOneB.h5")
testBoardState = {"red": [(-3,2),(0,-1),(-2,0),(1,1)],
"blue": [(-1,0),(1,-1),(-2,2)],
"green": [(1,-3),(0,0),(-3,1),(2,1)]}
testExitState = {"red": [(-1,-1),(-2,0),(-3,3),(-1,1)],
"blue": [],
"green": [(-2,1),(0,0)]}
testExits = {"red": 1, "blue": 0, "green": 0}
testCurrentPlayer = "red"
if False: # Switch this to test the eval function of the CNN
v, p = convNet.evaluate(boardState=testExitState, colour=testCurrentPlayer, exits=testExits)
nextMoves = getNextMoves(testExitState, testCurrentPlayer)
P = adjustNNPriors(p, testExitState, testCurrentPlayer)
for move in nextMoves:
print("{}: {}".format(move, P[encodeMove(move)]))
if False: # Switch this to test v outputs
full = [(-3,2),(0,-1),(-2,0),(1,1)]
for i in range(0, len(full)+1):
bs = deepcopy(testBoardState)
bs["red"] = full[0:i]
v, p = convNet.evaluate(boardState=bs, colour=testCurrentPlayer, exits=testExits)
clf = CNN(width=img_rows, height=img_columns, depth=1, total_classes=total_classes, weightsPath=args["weights"])
clf.compile(loss="categorical_crossentropy", optimizer=sgd, metrics=["accuracy"])
# Initially train and test the model; If weight saved already, load the weights using arguments.
b_size = 128 # Batch size
num_epoch = 20 # Number of epochs
verb = 1 # Verbose
# If weights saved and argument load_model; Load the pre-trained model.
if args["load_model"] < 0:
print('\nTraining the Model...')
clf.fit(trainData, trainLabels, batch_size=b_size, nb_epoch=num_epoch,verbose=verb)
# Evaluate accuracy and loss function of test data
print('Evaluating Accuracy and Loss Function...')
loss, accuracy = clf.evaluate(test_img, test_labels, batch_size=128, verbose=1)
print('Accuracy of Model: {:.2f}%'.format(accuracy * 100))
# Save the pre-trained model.
if args["save_model"] > 0:
print('Saving weights to file...')
clf.save_weights(args["weights"], overwrite=True)
# Show the images using OpenCV and making random selections.
for num in np.random.choice(np.arange(0, len(test_labels)), size=(5,)):
# Predict the label of digit using CNN.
probs = clf.predict(test_img[np.newaxis, i])
prediction = probs.argmax(axis=1)
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('float32')
x_test = x_test.astype('float32')
x_train /= 255
x_test /= 255
print('x_train shape:', x_train.shape, 'x_test.shape:', x_test.shape)
print(x_train.shape[0], 'train samples')
print(x_test.shape[0], 'test samples')
# Convert class vectors to binary class matrices
y_train = np_utils.to_categorical(y_train, num_classes)
y_test = np_utils.to_categorical(y_test, num_classes)
# Build the model
model = CNN(num_classes, input_shape, ngpus)
model.fit(x_train,
y_train,
batch_size=batch_size * ngpus,
epochs=nb_epochs,
verbose=1,
validation_data=(x_test, y_test))
score = model.evaluate(x_test, y_test, verbose=0)
print('Test score:', score[0])
print('Test accuracy:', score[1])
