def create_model(dim):
model = resnet18.make_resnet_18(dim).call(isTrain=False)
model.trainable = False
optimizer = tf.keras.optimizers.Adam(1e-4)
return model, optimizer
def create_model():
# base_model=tf.keras.applications.InceptionV3(
# include_top=False,
# weights='imagenet',
# input_tensor=None,
# input_shape=[227,227,3],
# pooling=None,
# classes=2
# )
# # tf.keras.utils.plot_model(base_model , to_file='model_base.png', show_shapes=True, dpi=64) #Added to visualise model
# base_model.trainable=False
# model = tf.keras.Sequential()
# model.add(base_model)
# model.add(tf.keras.layers.Flatten())
# model.add(tf.keras.layers.Dense(2))
# model.add(tf.keras.layers.Softmax())
model = resnet18.make_resnet_18(dim).call(isTrain=False)
model.trainable = False
optimizer = tf.keras.optimizers.Adam(1e-4)
return model, optimizer
train_summary_writer = tf.summary.create_file_writer(out_file)
# out_file_Test = STORE_PATH + f"/TEST_{dt.datetime.now().strftime('%d%m%Y%H%M')}"
# train_summary_writer_Test = tf.summary.create_file_writer(out_file_Test)
train_dir1 = 'H:/Datasets/celebA_Male_female/669126_1178231_bundle_archive/Dataset/Train/Male'
train_dir2 = 'H:/Datasets/celebA_Male_female/669126_1178231_bundle_archive/Dataset/Train/female'
# images_concat_shuffled,images_labels_concat_shuffled=data.prep_dataset(train_dir1,train_dir2)
images_labels_concat_shuffled, images_concat_shuffled = data.datasets(
train_dir1, train_dir2, 13000, bias=0.5)
# test_dir1 = 'H:/Datasets/celebA_Male_female/669126_1178231_bundle_archive/Dataset/Test/Male'
# test_dir2 = 'H:/Datasets/celebA_Male_female/669126_1178231_bundle_archive/Dataset/Test/female'
model = resnet18.make_resnet_18().call()
tf.keras.utils.plot_model(model,
to_file='resnet.png',
show_shapes=True,
dpi=64) #Added to visualise model
optimizer = tf.keras.optimizers.Adam(1e-4)
EPOCH = 10
BATCH_SIZE = 64
batch_num = len(images_labels_concat_shuffled) / BATCH_SIZE
def cross_entropy_loss(expected_labels, predicted_labels):
cross_entropy = tf.keras.losses.SparseCategoricalCrossentropy()
loss = cross_entropy(expected_labels, predicted_labels)