class Test(unittest.TestCase):
def setUp(self):
import Activity08_01
self.exercises = Activity08_01
self.file_url = 'https://github.com/PacktWorkshops/The-TensorFlow-Workshop/blob/master/Chapter08/dataset/fruits360.zip'
self.zip_dir = tf.keras.utils.get_file('fruits360.zip', origin=self.file_url, extract=True)
self.path = pathlib.Path(self.zip_dir).parent / 'fruits360_filtered'
self.train_dir = self.path / 'Training'
self.validation_dir = self.path / 'Test'
self.total_train = 11398
self.total_val = 4752
self.train_image_generator = ImageDataGenerator(rescale=1./255, rotation_range=40, width_shift_range=0.1, height_shift_range=0.1, shear_range=0.2, zoom_range=0.2, horizontal_flip=True, fill_mode='nearest')
self.validation_image_generator = ImageDataGenerator(rescale=1./255)
self.batch_size = 32
self.img_height = 224
self.img_width = 224
self.channel = 3
self.train_data_gen = self.train_image_generator.flow_from_directory(batch_size=self.batch_size, directory=self.train_dir, target_size=(self.img_height, self.img_width))
self.val_data_gen = self.validation_image_generator.flow_from_directory(batch_size=self.batch_size, directory=self.validation_dir, target_size=(self.img_height, self.img_width))
np.random.seed(8)
tf.random.set_seed(8)
self.base_model = NASNetMobile(include_top=False, input_shape=(self.img_height, self.img_width, self.channel), weights='imagenet')
for layer in self.base_model.layers[:700]:
layer.trainable = False
self.model = tf.keras.Sequential([
self.base_model,
layers.Flatten(),
layers.Dense(500, activation='relu'),
layers.Dense(120, activation='softmax')
])
self.model.compile(loss='categorical_crossentropy', optimizer=tf.keras.optimizers.Adam(0.001), metrics=['accuracy'])
def test_file_url(self):
self.assertEqual(self.exercises.file_url, self.file_url)
def test_total_train(self):
self.assertEqual(self.exercises.total_train, self.total_train)
def test_total_val(self):
self.assertEqual(self.exercises.total_val, self.total_val)
def test_base_model_summary(self):
self.assertEqual(self.exercises.base_model.summary(), self.base_model.summary())
def test_model_summary(self):
self.assertEqual(self.exercises.model.summary(), self.model.summary())
def build_model(self):
conv_base = NASNetMobile(include_top=False,
weights='imagenet',
input_shape=config.input_shape)
conv_base.trainable = False
print(conv_base.summary())
inputs = Input(shape=config.input_shape)
x = conv_base(inputs)
x = Flatten()(x)
x = Dense(units=256, activation=nasnet['activation'])(x)
outputs = Dense(units=config.num_classes,
activation=nasnet['outact'])(x)
self.model = Model(inputs=inputs, outputs=outputs)
aaa = 1
x_train = x_train.reshape(x_train.shape[0], x_train.shape[1], x_train.shape[2], aaa)
x_test = x_test.reshape(x_test.shape[0], x_test.shape[1], x_test.shape[2], aaa)
print(x_train.shape, y_train.shape) # (3628, 128, 862, 1) (3628,)
print(x_test.shape, y_test.shape) # (908, 128, 862, 1) (908,)
model = NASNetMobile(
include_top=True,
input_shape=(128,862,1),
classes=2,
pooling=None,
weights=None,
)
model.summary()
# model.trainable = False
model.save('C:/nmb/nmb_data/h5/5s/Nasnet/nasnet_adadelta_1.h5')
# 컴파일, 훈련
op = Adadelta(lr=1e-3)
batch_size = 4
es = EarlyStopping(monitor='val_loss', patience=20, restore_best_weights=True, verbose=1)
lr = ReduceLROnPlateau(monitor='val_loss', vactor=0.5, patience=10, verbose=1)
path = 'C:/nmb/nmb_data/h5/5s/Nasnet/nasnet_adadelta_1.h5'
mc = ModelCheckpoint(path, monitor='val_loss', verbose=1, save_best_only=True)
model.compile(optimizer=op, loss="sparse_categorical_crossentropy", metrics=['acc'])
history = model.fit(x_train, y_train, epochs=1000, batch_size=batch_size, validation_split=0.2, callbacks=[es, lr, mc])
from tensorflow.keras.layers import Dense, Flatten, UpSampling2D
from tensorflow.keras.callbacks import EarlyStopping, ReduceLROnPlateau
(x_train, y_train), (x_test, y_test) = cifar10.load_data()
y_train = to_categorical(y_train)
y_test = to_categorical(y_test)
print(x_train.shape, x_test.shape)
print(y_train.shape, y_test.shape)
nasnetmobile = NASNetMobile(weights='imagenet',
include_top=False,
input_shape=(224, 224, 3))
nasnetmobile.summary()
nasnetmobile.trainable = False
model = Sequential()
model.add(UpSampling2D(size=(7, 7)))
model.add(nasnetmobile)
model.add(Flatten())
model.add(Dense(256))
model.add(Dense(64))
model.add(Dense(10, activation='softmax'))
# model.summary()
model.compile(loss='categorical_crossentropy',
optimizer='adam',
# test_dataset = train.flow_from_directory('C:/Users/param/Face-Mask-Detection/dataset/test',
# target_size=(224, 224),
# batch_size=32,
# class_mode='binary')
# train_dataset.class_indices
# local_weights_file = 'C:/Users/param/Face-Mask-Detection/face_detector/inceptionv3-model-10ep.h5'
pre_trained_model = NASNetMobile(weights="imagenet",
include_top=False,
input_tensor=Input(shape=(224, 224, 3)))
# pre_trained_model.load_weights(local_weights_file)
#
#
pre_trained_model.summary()
last_layer = pre_trained_model.get_layer('mixed7')
# print('last layer output shape: ', last_layer.output_shape)
last_output = last_layer.output
# construct the head of the model that will be placed on top of the
# the base model
# headModel = pre_trained_model.output
# headModel = AveragePooling2D(pool_size=(5, 5))(last_output)
# headModel = BatchNormalization(axis=1)(last_output)
# headModel = Conv2D(32, (3, 3), padding="same", activation="relu")(headModel)
headModel = AveragePooling2D(pool_size=(5, 5))(last_output)
headModel = Flatten(name="flatten")(headModel)
# headModel = Dropout(0.5)(headModel)
headModel = Dense(512, activation="relu")(headModel)
headModel = Dropout(0.5)(headModel)
class Test(unittest.TestCase):
def setUp(self):
import Exercise08_01
self.exercises = Exercise08_01
self.file_url = 'https://storage.googleapis.com/mledu-datasets/cats_and_dogs_filtered.zip'
self.zip_dir = tf.keras.utils.get_file('cats_and_dogs.zip',
origin=self.file_url,
extract=True)
self.path = pathlib.Path(
self.zip_dir).parent / 'cats_and_dogs_filtered'
self.train_dir = self.path / 'train'
self.validation_dir = self.path / 'validation'
self.total_train = 2000
self.total_val = 1000
self.train_image_generator = ImageDataGenerator(rescale=1. / 255)
self.validation_image_generator = ImageDataGenerator(rescale=1. / 255)
self.batch_size = 32
self.img_height = 224
self.img_width = 224
self.channel = 3
self.train_data_gen = self.train_image_generator.flow_from_directory(
batch_size=self.batch_size,
directory=self.train_dir,
shuffle=True,
target_size=(self.img_height, self.img_width),
class_mode='binary')
self.val_data_gen = self.validation_image_generator.flow_from_directory(
batch_size=self.batch_size,
directory=self.validation_dir,
target_size=(self.img_height, self.img_width),
class_mode='binary')
np.random.seed(8)
tf.random.set_seed(8)
self.base_model = NASNetMobile(include_top=False,
input_shape=(self.img_height,
self.img_width,
self.channel),
weights='imagenet')
self.base_model.trainable = False
self.model = tf.keras.Sequential([
self.base_model,
layers.Flatten(),
layers.Dense(500, activation='relu'),
layers.Dense(1, activation='sigmoid')
])
self.model.compile(loss='binary_crossentropy',
optimizer=tf.keras.optimizers.Adam(0.001),
metrics=['accuracy'])
def test_file_url(self):
self.assertEqual(self.exercises.file_url, self.file_url)
def test_total_train(self):
self.assertEqual(self.exercises.total_train, self.total_train)
def test_total_val(self):
self.assertEqual(self.exercises.total_val, self.total_val)
def test_base_model_summary(self):
self.assertEqual(self.exercises.base_model.summary(),
self.base_model.summary())
def test_model_summary(self):
self.assertEqual(self.exercises.model.summary(), self.model.summary())