def test_append_dense_layer():
model = CNN()
model.add_input_layer(shape=(256 * 256 * 3), name="input0")
model.append_dense_layer(num_nodes=100, activation='relu')
input = np.zeros((10, 256 * 256 * 3))
result = model.predict(input)
assert result.shape == (10, 100)
def test_add_flatten_layer():
model = CNN()
model.add_input_layer(shape=(256, 256, 3), name="input0")
model.append_flatten_layer(name='flatten')
input = np.zeros((10, 256, 256, 3))
out = model.predict(input)
assert out.shape == (10, 256 * 256 * 3)
def test_append_conv2d_layer():
model = CNN()
model.add_input_layer(shape=(256, 256, 3), name="input0")
model.append_conv2d_layer(10, (3, 3), activation='relu')
input = np.zeros((20, 256, 256, 3))
out = model.predict(input)
assert (out.shape == (20, 256, 256, 10))
def test_append_maxpooling2d_layer():
model = CNN()
model.add_input_layer(shape=(256, 256, 3), name="input0")
model.append_maxpooling2d_layer(pool_size=(2, 2),
padding="same",
strides=2,
name='pooling')
input = np.zeros((10, 256, 256, 3))
out = model.predict(input)
assert (out.shape == (10, 128, 128, 3))
def test_remove_last_layer():
from tensorflow.keras.datasets import cifar10
batch_size = 32
num_classes = 10
epochs = 100
data_augmentation = True
num_predictions = 20
save_dir = os.path.join(os.getcwd(), 'saved_models')
model_name = 'keras_cifar10_trained_model.h5'
(X_train, y_train), (X_test, y_test) = cifar10.load_data()
number_of_train_samples_to_use = 100
X_train = X_train[0:number_of_train_samples_to_use, :]
y_train = y_train[0:number_of_train_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=16,
kernel_size=(3, 3),
padding="same",
activation='linear',
name="conv1")
my_cnn.append_maxpooling2d_layer(pool_size=2,
padding="same",
strides=2,
name="pool1")
my_cnn.append_conv2d_layer(num_of_filters=8,
kernel_size=3,
activation='relu',
name="conv2")
my_cnn.append_flatten_layer(name="flat1")
my_cnn.append_dense_layer(num_nodes=10, activation="relu", name="dense1")
my_cnn.append_dense_layer(num_nodes=2, activation="relu", name="dense2")
out = my_cnn.predict(X_train)
assert out.shape == (number_of_train_samples_to_use, 2)
my_cnn.remove_last_layer()
out = my_cnn.predict(X_train)
assert out.shape == (number_of_train_samples_to_use, 10)
def test_predict():
# some of these may be duplicated
X = np.float32([[0.1, 0.2, 0.3, 0.4, 0.5, -0.1, -0.2, -0.3, -0.4, -0.5]])
X = np.float32([[0.1, 0.2, 0.3, 0.4, 0.5, 0, 0, 0, 0, 0]])
X = np.float32([np.linspace(0, 10, num=10)])
# X = np.float32([[0.1, 0.2]])
my_cnn = CNN()
my_cnn.add_input_layer(shape=(10, ), name="input0")
my_cnn.append_dense_layer(num_nodes=5, activation='linear', name="layer1")
w = my_cnn.get_weights_without_biases(layer_name="layer1")
w_set = np.full_like(w, 2)
my_cnn.set_weights_without_biases(w_set, layer_name="layer1")
b = my_cnn.get_biases(layer_name="layer1")
b_set = np.full_like(b, 2)
b_set[0] = b_set[0] * 2
my_cnn.set_biases(b_set, layer_name="layer1")
# my_cnn.append_dense_layer(num_nodes=5, activation='linear', name="layer12")
actual = my_cnn.predict(X)
assert np.array_equal(actual, np.array([[104., 102., 102., 102., 102.]]))