def network(inputs):
previous = resnet50.ResNet50(
include_top=False,
input_tensor=inputs,
pooling=pooling,
weights=weights,
).output
return tf.keras.layers.Flatten(name="flatten")(previous)
def __init__(self, settings: Settings, view_id, *args, **kwargs):
super().__init__(*args, **kwargs)
self.settings = settings
self.view_id = view_id
# region Use VGG16
if self.settings.svcnn_model == 'vgg16':
vgg16_model = vgg16.VGG16(include_top=False,
input_shape=settings.input_shape,
pooling=None,
weights=None)
x = vgg16_model.layers[-1].output
x = Flatten(name='flatten')(x)
x = Dense(4096, activation='relu', name='fc1')(x)
x = Dense(4096, activation='relu', name='fc2')(x)
x = Dense(settings.num_classes,
activation='softmax',
name='predictions')(x)
self.model_input = vgg16_model.input
self.model_output = x
self.model = Model(inputs=self.model_input,
outputs=self.model_output)
# endregion
# region Use VGG19
elif self.settings.svcnn_model == 'vgg19':
vgg19_model = vgg19.VGG19(include_top=False,
input_shape=settings.input_shape,
pooling=None,
weights=None)
x = vgg19_model.layers[-1].output
x = Flatten(name='flatten')(x)
x = Dense(4096, activation='relu', name='fc1')(x)
x = Dense(4096, activation='relu', name='fc2')(x)
x = Dense(settings.num_classes,
activation='softmax',
name='predictions')(x)
self.model_input = vgg19_model.input
self.model_output = x
self.model = Model(inputs=self.model_input,
outputs=self.model_output)
# endregion
# region Use ResNet50
elif self.settings.svcnn_model == 'resnet50':
resnet50_model = resnet50.ResNet50(
include_top=False,
input_shape=settings.input_shape,
pooling=None,
weights=None)
x = resnet50_model.layers[-1].output
x = GlobalAveragePooling2D(name='avg_pool')(x)
x = Dense(settings.num_classes, activation='softmax',
name='probs')(x)
self.model_input = resnet50_model.input
self.model_output = x
self.model = Model(inputs=self.model_input,
outputs=self.model_output)
def build_model(input_shape, output_shape):
pretrained_model = resnet50.ResNet50(weights='imagenet', include_top=False)
x = pretrained_model.output
x = GlobalAveragePooling2D()(x)
x = Dense(2048, activation='relu')(x)
x = Dropout(0.5)(x)
outputs = Dense(output_shape, activation='softmax')(x)
model = Model(inputs=pretrained_model.input, outputs=outputs)
return model
def __init__(self,
num_classes,
scales=[0.05, 0.1, 0.2, 0.4, 0.6, 0.8],
base_layers=[
'conv4_block4_out', 'conv4_block5_out',
'conv4_block6_out', 'conv5_block1_out',
'conv5_block2_out', 'conv5_block3_out'
],
aspect_ratios=[0.5, 1.0, 2.0],
variances=[0.1, 0.1, 0.2, 0.2],
weight_decay=5e-4,
**kwargs):
super().__init__(**kwargs)
self.num_classes = num_classes
if len(scales) != len(base_layers):
raise Exception(
'You need to provide one scale for each base layer.')
self.scales = scales
self.base_layers = base_layers
self.aspect_ratios = aspect_ratios
self.boxes_per_cell = len(aspect_ratios)
if len(variances) != 4:
raise Exception('You need to provide exactly 4 variance values \
(one for each bounding box parameter).')
self.variances = variances
backbone = resnet50.ResNet50(include_top=False, weights='imagenet')
self.get_base_features = tf.keras.Model(
inputs=backbone.layers[0].input,
outputs=[
backbone.get_layer(layer_name).output
for layer_name in self.base_layers
])
self.conv_cls = []
self.conv_loc = []
for idx in range(len(self.scales)):
self.conv_cls.append(
layers.Conv2D(
filters=self.boxes_per_cell * self.num_classes,
kernel_size=3,
padding='same',
kernel_regularizer=tf.keras.regularizers.l2(weight_decay),
name='conv_cls_{}'.format(idx + 1)))
self.conv_loc.append(
layers.Conv2D(
filters=self.boxes_per_cell * 4,
kernel_size=3,
padding='same',
kernel_regularizer=tf.keras.regularizers.l2(weight_decay),
name='conv_loc_{}'.format(idx + 1)))
def ResNetSegmentation(n_classes, input_height=224, input_width=224):
img_input = Input(shape=(input_height, input_width, 3))
resnet_model = resnet50.ResNet50(input_tensor=img_input,
weights="imagenet")
o = resnet_model.get_layer('activation_39').output
o = (ZeroPadding2D((1, 1), data_format='channels_last'))(o)
o = (Conv2D(512, (3, 3),
padding='valid',
data_format='channels_last',
kernel_initializer='he_uniform'))(o)
o = (BatchNormalization())(o)
o = (UpSampling2D((2, 2), data_format='channels_last'))(o)
o = (ZeroPadding2D((1, 1), data_format='channels_last'))(o)
o = (Conv2D(512, (3, 3),
padding='valid',
data_format='channels_last',
kernel_initializer='he_uniform'))(o)
o = (BatchNormalization())(o)
o = (UpSampling2D((2, 2), data_format='channels_last'))(o)
o = (ZeroPadding2D((1, 1), data_format='channels_last'))(o)
o = (Conv2D(256, (3, 3),
padding='valid',
data_format='channels_last',
kernel_initializer='he_uniform'))(o)
o = (BatchNormalization())(o)
o = (UpSampling2D((2, 2), data_format='channels_last'))(o)
o = (ZeroPadding2D((1, 1), data_format='channels_last'))(o)
o = (Conv2D(128, (3, 3),
padding='valid',
data_format='channels_last',
kernel_initializer='he_uniform'))(o)
o = (BatchNormalization())(o)
o = (UpSampling2D((2, 2), data_format='channels_last'))(o)
o = (ZeroPadding2D((1, 1), data_format='channels_last'))(o)
o = (Conv2D(64, (3, 3),
padding='valid',
data_format='channels_last',
kernel_initializer='he_uniform'))(o)
o = (BatchNormalization())(o)
o = Conv2D(n_classes, (3, 3),
padding='same',
data_format='channels_last',
kernel_initializer='he_uniform')(o)
o = (Activation('softmax'))(o)
model = Model(img_input, o)
return model
def simple_model(config):
model = resnet50.ResNet50(weights=None)
# model = Sequential([Dense(10, input_shape=(1, )), Dense(1)])
model.compile(optimizer="Adam",
loss="categorical_crossentropy",
metrics=["categorical_crossentropy"])
return model
def get_resnet50(img_shape=(80,80,3)):
'''
Arguments: input image shape
default = (80,80)
'''
image_input = keras.Input(shape=img_shape) #Define the input layer in our embedding_extraction_model
temp_model = resnet50.ResNet50(weights='imagenet',include_top=False)(image_input) #Set resnet50 to temp_model
image_embedding_output_layer = keras.layers.GlobalAveragePooling2D()(temp_model)
img_embedding_extractor_model = Model(inputs=image_input, outputs=image_embedding_output_layer) #Create a new model and add GlobalAveragePooling2D layer to it.
return img_embedding_extractor_model
def __init__(self, classes=6, units=512, dropout_prob=0.5, l1=0.01, l2=0.01):
self.base_model = resnet50.ResNet50(
weights = 'imagenet',
input_shape=(224,224,3),
include_top = False,
pooling = 'avg')
x = Dense(units, activation='relu')(self.base_model.output)
x = Dropout(dropout_prob)(x)
y = Dense(classes, activation='softmax', kernel_regularizer=l1_l2(l1=l1, l2=l2))(x)
self.model = Model(inputs=self.base_model.input, outputs=y)
def __init__(self, base_path='.'):
"""
Initializes main variables
@param base_path: string pointing to the path where the images are located. If no string is indicated, the
current directory will be considered
"""
super(ResNet50FeatureExtractor, self).__init__(base_path)
self.model = resnet50.ResNet50(weights='imagenet',
include_top=False,
pooling='avg')
def __init__(self, latent_dim, input_shape, rotation_ranges):
super(RealEncoder, self).__init__()
self.resnet = resnet50.ResNet50(weights="imagenet", include_top=False, input_shape=input_shape, pooling="avg")
self.resnet_feature_dim = np.prod(self.resnet.output.shape[1:])
self.rotation_regressor = keras.layers.Dense(3, input_shape=(self.resnet_feature_dim,), activation=tf.nn.tanh)
self.feature_to_latent_mlp = keras.layers.Dense(latent_dim, input_shape=(self.resnet_feature_dim,))
self.rotation_range_multiplier = np.array([rotation_ranges[0][1], rotation_ranges[1][1], rotation_ranges[2][1]])
self.rotation_range_multiplier = np.pi * self.rotation_range_multiplier / 180.0
def build_image_model() -> tf.keras.Model:
preprocess_input = resnet50.preprocess_input
resnet_model = resnet50.ResNet50(weights="imagenet", include_top=False)
global_max_pooling = tf.keras.layers.GlobalAveragePooling2D()
image_inputs = tf.keras.Input(shape=IMAGE_NET_IMAGE_SIZE +
(IMAGE_NET_CHANNELS, ),
name="image")
x = preprocess_input(image_inputs)
x = resnet_model(x, training=False)
image_embedding = global_max_pooling(x)
return tf.keras.Model(image_inputs, image_embedding, name="ResNet50")
def createResNetwork():
model = resnet50.ResNet50(weights='imagenet')
for layer in model.layers:
layer._name = 'res_'+layer._name
x = Flatten()(model.get_layer('res_avg_pool').output)
x = Dropout(0.3)(x)
x = Dense(256, name='res_weights', activation=LeakyReLU(alpha=0.3))(x)
x = Dense(4, activation='softmax')(x)
model = Model(model.input, x)
return model
def init():
# Instantiate Resnet50 featurizer
global featurizer
featurizer = resnet50.ResNet50(weights='imagenet',
input_shape=(224, 224, 3),
include_top=False,
pooling='avg')
# Load the model
global model
# retreive the path to the model file using the model name
model_path = Model.get_model_path('aerial_classifier')
model = tf.keras.models.load_model(model_path)
def init():
# Instantiate ResNet50 featurizer
global featurizer
featurizer = resnet50.ResNet50(weights='imagenet',
input_shape=(224, 224, 3),
include_top=False,
pooling='avg')
# Load the model
global model
# retreive the path to the model file using the model name
model_path = Model.get_model_path(model_name='AutoMLac5fd8f37best')
model = joblib.load(model_path)
def __init__(self):
model_path = '../model/resnet50_weights_tf_dim_ordering_tf_kernels_notop.h5'
if path.exists(model_path) == False:
print('Downloading the model weights...')
url = 'https://github.com/keras-team/keras-applications/releases/download/resnet/resnet50_weights_tf_dim_ordering_tf_kernels_notop.h5'
with requests.get(url, stream=True) as r:
with open(model_path, 'wb') as f:
shutil.copyfileobj(r.raw, f)
print('Loading the model weights from \n', model_path)
model = resnet50.ResNet50(include_top=False, weights=model_path, pooling='avg')
self.embedding_model = Model(inputs=model.input,
outputs=model.output)
def _get_model(self, model_name):
if model_name == 'resnet':
self.preprocess_input = resnet50.preprocess_input
base_model = resnet50.ResNet50(include_top=False,
input_shape=self.target_shape)
for layer in base_model.layers:
layer.trainable = False
model = Sequential([base_model, GlobalAveragePooling2D()])
return model
return None
def __init__(self, model_name='resnet'):
if (model_name == 'mobilenet'):
print("Using MobileNetV2 to vectorize images")
model = mobilenet_v2.MobileNetV2(weights='imagenet')
layer_name = 'global_average_pooling2d'
self.preprocess_fn = mobilenet_v2.preprocess_input
else:
print("Using ResNet50 to vectorize images")
model = resnet50.ResNet50(weights='imagenet')
layer_name = 'avg_pool'
self.preprocess_fn = resnet50.preprocess_input
o = model.get_layer(layer_name).output
self.vec_len = o.shape[1]
self.intermediate_layer_model = Model(inputs=model.input,
outputs=o)
def compute_repr_resnet(self, x_train: np.ndarray, startTime: float):
from tensorflow.keras.applications import resnet50
from tensorflow import stack, concat, squeeze
from tensorflow.compat.v1.image import resize_images
from tensorflow.keras.models import Model
# resize function
target_size = 224
def encode_samples(x_sample):
batch_size = 1000
no_samples = x_sample.shape[0]
index = 0
x_features = []
while index < no_samples:
print("encoding sample %d at time %f" %
(index, (time.time() - startTime)))
end = min(index + batch_size, no_samples)
x = x_sample[index:end]
if x.shape[
-1] == 1: # add 2 dimensions to make rgb sample out of black-white-sample
x = concat((x, ) * 3, axis=-1)
x = squeeze(x)
x = resize_images(x, size=(target_size, target_size))
x = resnet50.preprocess_input(x)
x_feat = intermediate_layer_model.predict(x, workers=16)
x_features += [x_feat]
index += batch_size
x_feat = concat(x_features, axis=0)
return x_feat
# encode function
model = resnet50.ResNet50(weights='imagenet',
input_shape=(target_size, target_size, 3))
layer_name = 'avg_pool'
intermediate_layer_model = Model(
inputs=model.inputs, outputs=model.get_layer(layer_name).output)
# intermediate_layer_model.compile(loss="mse",optimizer="sgd")
def encode(x_samples):
encoded_samples = encode_samples(x_samples)
return encoded_samples
# encode training data
x_train_repr_1d = encode(x_train)
return x_train_repr_1d
def create_bottleneck_features():
# Configure input directories
train_images_dir = os.path.join(FLAGS.input_data_dir, 'train')
valid_images_dir = os.path.join(FLAGS.input_data_dir, 'valid')
# Create generators for training and validation data
train_generator = ImageGenerator(train_images_dir,
resnet50.preprocess_input)
valid_generator = ImageGenerator(valid_images_dir,
resnet50.preprocess_input)
# Create a featurizer
featurizer = resnet50.ResNet50(weights='imagenet',
input_shape=(224, 224, 3),
include_top=False,
pooling='avg')
# Generate training bottleneck features
print("Generating training bottleneck features")
features = featurizer.predict_generator(train_generator, verbose=1)
labels = train_generator.get_labels()
# Save training dataset to HDF5 file
filename = 'aerial_bottleneck_train.h5'
output_file = os.path.join(FLAGS.output_data_dir, filename)
print("Saving training features to {}".format(output_file))
print(" Training features: ", features.shape)
print(" Training labels: ", labels.shape)
with h5py.File(output_file, "w") as hfile:
features_dset = hfile.create_dataset('features', data=features)
labels_dset = hfile.create_dataset('labels', data=labels)
# Generate validation bottleneck features
print("Generating validation bottleneck features")
features = featurizer.predict_generator(valid_generator, verbose=1)
labels = valid_generator.get_labels()
# Save validation dataset to HDF5 file
filename = 'aerial_bottleneck_valid.h5'
output_file = os.path.join(FLAGS.output_data_dir, filename)
print("Saving validation features to {}".format(output_file))
print(" Validation features: ", features.shape)
print(" Validation labels: ", labels.shape)
with h5py.File(output_file, "w") as hfile:
features_dset = hfile.create_dataset('features', data=features)
labels_dset = hfile.create_dataset('labels', data=labels)
print("Done")
def build_model():
input_tensor_shape = (None, 224, 224, 3)
input_img = tf.compat.v1.placeholder(tf.float32,
input_tensor_shape,
name='input_img')
x = resnet50.preprocess_input(input_img)
resnet = resnet50.ResNet50(weights='imagenet',
include_top=False,
pooling='avg')
output_tensor = resnet(x)
return input_img, output_tensor
def predict(fname):
input_shape = (224, 224, 3)
# load and resize image
img = image.load_img(fname, target_size=input_shape[:2])
x = image.img_to_array(img)
# preprocess image
# make a batch
import numpy as np
x = np.expand_dims(x, axis=0)
print(x.shape)
# apply the preprocessing function of resnet50
img_array = resnet50.preprocess_input(x)
model = resnet50.ResNet50(weights='imagenet', input_shape=input_shape)
preds = model.predict(x)
return resnet50.decode_predictions(preds)
def init():
try:
# Create ResNet50 featurizer
global featurizer
featurizer = resnet50.ResNet50(weights='imagenet',
input_shape=(224, 224, 3),
include_top=False,
pooling='avg')
# Load top model
global model
model_name = 'aerial_classifier'
model_path = Model.get_model_path(model_name)
model = tf.keras.models.load_model(model_path)
except Exception as e:
print('Exception during init: ', str(e))
def unet_resnet(input_shape):
resnet_base = resnet50.ResNet50(input_shape=input_shape, include_top=False)
for layer in resnet_base.layers:
layer.trainable = True
conv1 = resnet_base.get_layer("activation").output
conv2 = resnet_base.get_layer("activation_9").output
conv3 = resnet_base.get_layer("activation_21").output
conv4 = resnet_base.get_layer("activation_39").output
conv5 = resnet_base.get_layer("activation_48").output
conv6 = _merge_block(conv5, conv4, 256)
conv7 = _merge_block(conv6, conv3, 192)
conv8 = _merge_block(conv7, conv2, 128)
conv9 = _merge_block(conv8, conv1, 64)
conv10 = _merge_block(conv9, resnet_base.input, 32)
x = _end_block(conv10, spatial_dropout=0.25)
model = Model(resnet_base.input, x)
return model, "resnet"
def train_evaluate(run):
# Create bottleneck featurs
train_images_dir = os.path.join(FLAGS.data_folder, 'train')
valid_images_dir = os.path.join(FLAGS.data_folder, 'valid')
train_generator = ImageGenerator(train_images_dir, resnet50.preprocess_input)
valid_generator = ImageGenerator(valid_images_dir, resnet50.preprocess_input)
featurizer = resnet50.ResNet50(
weights = 'imagenet',
input_shape=(224,224,3),
include_top = False,
pooling = 'avg')
print("Generating bottleneck features")
train_features = featurizer.predict_generator(train_generator, verbose=1)
train_labels = train_generator.get_labels()
valid_features = featurizer.predict_generator(valid_generator, verbose=1)
valid_labels = valid_generator.get_labels()
# Create a classifier
model = fcn_classifier(input_shape=(2048,), units=FLAGS.units, l1=FLAGS.l1, l2=FLAGS.l2)
# Start training
print("Starting training")
model.fit(train_features, train_labels,
batch_size=64,
epochs=20,
shuffle=True,
validation_data=(valid_features, valid_labels))
# Save the trained model to outp'uts which is a standard folder expected by AML
print("Training completed.")
os.makedirs('outputs', exist_ok=True)
model_file = os.path.join(FLAGS.save_model_dir, 'model.hd5')
print("Saving model to: {0}".format(model_file))
model.save(model_file)
def get_model(model_name):
if model_name == "vgg16":
return vgg16.VGG16(weights='imagenet'
), vgg16.decode_predictions, vgg16.preprocess_input
elif model_name == "vgg19":
return vgg19.VGG19(weights='imagenet'
), vgg19.decode_predictions, vgg19.preprocess_input
elif model_name == "resnet50":
return resnet50.ResNet50(
weights='imagenet'
), resnet50.decode_predictions, resnet50.preprocess_input
elif model_name == "resnet101":
return ResNet101(weights='imagenet'
), resnet.decode_predictions, resnet.preprocess_input
elif model_name == "mobilenet":
return mobilenet.MobileNet(
weights='imagenet'
), mobilenet.decode_predictions, mobilenet.preprocess_input
elif model_name == "densenet":
return densenet.DenseNet121(
weights='imagenet'
), densenet.decode_predictions, densenet.preprocess_input
def custom_model():
base_model = resnet50.ResNet50(input_shape=(224, 224, 3),
include_top=False,
weights='imagenet',
pooling='avg')
x = base_model.output
x = keras.layers.Dense(1024, activation='relu')(x)
x = keras.layers.Dropout(0.5)(x)
out = keras.layers.Dense(2, activation='softmax', name='output_layer')(x)
custom_resnet_model = keras.models.Model(inputs=base_model.input,
outputs=out)
for layer in base_model.layers:
layer.trainable = False
custom_resnet_model.compile(
optimizer=tf.keras.optimizers.Adam(),
loss=tf.keras.losses.sparse_categorical_crossentropy,
metrics=['accuracy'])
return custom_resnet_model
import numpy as np
from tensorflow.keras.preprocessing import image
from tensorflow.keras.applications import resnet50
# Load Keras' ResNet50 model that was pre-trained against the ImageNet database
model = resnet50.ResNet50()
# Load the image file, resizing it to 224x224 pixels (required by this model)
# img = image.load_img("bay.jpg", target_size=(224,224))
# img = image.load_img("bonk.jpg", target_size=(224,224))
img = image.load_img("froggy.jpg", target_size=(224, 224))
# Convert the image to a numpy array
x = image.img_to_array(img)
# Add a forth dimension since Keras expects a list of images
x = np.expand_dims(x, axis=0)
# Scale the input image to the range used in the trained network
x = resnet50.preprocess_input(x)
# Run the image through the deep neural network to make a prediction
predictions = model.predict(x)
# Look up the names of the predicted classes. Index zero is the results for the first image.
predicted_classes = resnet50.decode_predictions(predictions, top=10)
print("This is an image of:")
for imagenet_id, name, likelihood in predicted_classes[0]:
print(" - {}: {:2f} likelihood".format(name, likelihood))
""" trains the model and save it """
batch_size = 32
num_classes = 10
epochs = 20
(x_train, y_train), (x_test, y_test) = K.datasets.cifar10.load_data()
x_train, y_train = preprocess_data(x_train, y_train)
x_test, y_test = preprocess_data(x_test, y_test)
"""Setting up the model"""
original_dim = (32, 32, 3)
target_size = (224, 224)
resN = resnet50.ResNet50(include_top=False, weights='imagenet',
input_shape=original_dim,
pooling='max')
# Defrost last layer of resnet50
for layer in resN.layers[:-32]:
layer.trainable = False
# Concatenate the resnet with a modified top layer
res_model = K.Sequential()
res_model.add(K.layers.Lambda(lambda image:
tf.image.resize(image, target_size)))
res_model.add(resN)
res_model.add(K.layers.Dense(512, activation='relu'))
res_model.add(K.layers.Dropout(0.3))
res_model.add(K.layers.Dense(512, activation='relu'))
res_model.add(K.layers.Dropout(0.5))
res_model.add(K.layers.Dense(10, activation='softmax'))
image_size = tuple((224, 224))
image_bytes = cv2.resize(image_bytes, image_size)
feature_table = {'resnet': None}
# feature_table['vgg'] = vgg(image_bytes)
feature_table['resnet'] = Resnet50(image_bytes)
return feature_table
# def cluster():
# kmeans = KMeans(n_clusters=2, random_state=0).fit(array)
if __name__ == "__main__":
# vgg_model = tf.keras.applications.VGG16(weights='imagenet')
# vgg_extractor = tf.keras.models.Model(inputs=vgg_model.input, outputs=vgg_model.get_layer("fc2").output)
resnet50_extractor = resnet50.ResNet50(weights='imagenet',
include_top=False,
input_shape=(224, 224, 3))
img_dir_path = input('[INPUT] image dir path : ')
features = {'img': [], 'resnet': [], 'cluster': []}
pics_num = os.listdir(img_dir_path)
bar = progressbar.ProgressBar(maxval=len(pics_num), \
widgets=[progressbar.Bar('=', '[', ']'), ' ', progressbar.Percentage()])
bar.start()
for i, img_path in enumerate(pics_num):
img_path = img_dir_path + img_path
with open(img_path, 'rb') as f:
img_bytes = f.read()
Image = cv2.imdecode(np.fromstring(img_bytes, np.uint8),
cv2.IMREAD_UNCHANGED)
Image = Image[:, :, :3]
single_feature_table = feature_table_creator(Image)
from tensorflow.keras import models
from tensorflow.keras import optimizers
from tensorflow.keras.applications import vgg16, inception_v3, resnet50, mobilenet
from tensorflow.keras.preprocessing.image import ImageDataGenerator
from tensorflow.keras.preprocessing.image import img_to_array, load_img
from tensorflow.keras.callbacks import TensorBoard
### ImageNet Large Scale Visual Recognition Challenge (ILSVRC) (1.2 M images, 1000 classes)
# Load the VGG model
vgg_model = vgg16.VGG16(weights='imagenet')
# Load the inception_V3 model
inception_model = inception_v3.InceptionV3(weights='imagenet')
# Load the ResNet50 model
resnet_model = resnet50.ResNet50(weights='imagenet')
# Load the MobileNet model
mobilenet_model = mobilenet.MobileNet(weights='imagenet')
### Load the image and convert its format to a 4-dimensional Tensor as an input of the form
# (batchsize, height, width, channels) requested by the Network.
filename = 'C:/Users/Theo/PycharmProjects/BMDATA/TP1/data/train/cat.1.jpg'
# Load an image in a PIL format
original = load_img(filename, target_size=(224, 224))
numpy_image = img_to_array(original)
# We add the extra dimension to the axis 0
image_batch = np.expand_dims(numpy_image, axis=0)
print('image batch size', image_batch.shape)
plt.imshow(np.uint8(image_batch[0]))