print(f"Dataset: Train: {len(x_train)} - Valid: {len(x_valid)} - Test: {len(x_test)}")
""" Hyperparameters """
shape = (2048,1024,3)
num_classes = 30
lr = 1e-4
bacth_size = 8
epochs = 10
""" Model """
model = build_Unet(shape, num_classes)
model.compile(loss = "categorical_crossentropy", optimizer=tf.keras.optimizers.Adam(lr),metrics=["accuracy"])
model.summary()
""" data """
train_dataset = tf_dataset(x_train, y_train, batch=bacth_size)
valid_dataset = tf_dataset(x_valid, y_valid, batch=bacth_size)
train_steps = len(x_train)//bacth_size
valid_steps = len(x_valid)//bacth_size
callbacks = [
ModelCheckpoint("model.h5",verbose=1, save_best_only=True),
ReduceLROnPlateau(monitor="val_loss",patience=1,factor=0.1, verbose=1,min_lr=1e-6),
EarlyStopping(monitor="val_loss", patience=5,verbose=1)
]
start = datetime.now()
model.fit( train_dataset,
steps_per_epoch=train_steps,
validation_steps=valid_steps,
validation_data=valid_dataset,
]
mask = np.concatenate(all_images, axis=1)
cv2.imwrite(f"results/{name}", mask)
if __name__ == "__main__":
print("")
## Seeding
np.random.seed(42)
tf.random.set_seed(42)
## Creating folders
create_dir("results/")
## Hyperparameters
batch_size = 32
test_path = "../new_data/test/"
test_x = sorted(glob(os.path.join(test_path, "image", "*.jpg")))
test_y = sorted(glob(os.path.join(test_path, "mask", "*.jpg")))
test_dataset = tf_dataset(test_x, test_y, batch=batch_size)
test_steps = (len(test_x) // batch_size)
if len(test_x) % batch_size != 0:
test_steps += 1
model = load_model("files/model.h5")
model.evaluate(test_dataset, steps=test_steps)
evaluate_normal(model, test_x, test_y)
recall = true_positives / (possible_positives + K.epsilon())
f1_val = 2*(precision*recall)/(precision+recall+K.epsilon())
return f1_val
if __name__ == "__main__":
## Dataset
path = "new/"
(train_x, train_y), (valid_x, valid_y), (test_x, test_y) = load_data(path)
## Hyperparameters
batch = 32
lr = 1e-4
epochs = 20
train_dataset = tf_dataset(train_x, train_y, batch=batch)
valid_dataset = tf_dataset(valid_x, valid_y, batch=batch)
model = build_model()
opt = tf.keras.optimizers.Adam(lr)
metrics = ["acc", tf.keras.metrics.Recall(), tf.keras.metrics.Precision(), f1_score]
model.compile(loss="binary_crossentropy", optimizer=opt, metrics=metrics)
tensorboard_callback = tf.keras.callbacks.TensorBoard(log_dir="./logs")
callbacks = [
ModelCheckpoint("files/model.h5"),
ReduceLROnPlateau(monitor='val_loss', factor=0.1, patience=4),
CSVLogger("files/data.csv"),
tensorboard_callback,
EarlyStopping(monitor='val_loss', patience=10, restore_best_weights=False)
import matplotlib.pyplot as plt
if __name__ == "__main__":
""" Hyperparamaters """
dataset_path = "C:/Users/VISWKADE/Documents/Uniklinik_a/Uniklinik_a/1PIC/*"
input_shape = (256, 256, 3)
batch_size = 12
epochs = 100
lr = 1e-4
model_path = "unet.h5"
csv_path = "data.csv"
""" Load the dataset """
(train_x, train_y), (test_x, test_y) = load_dataset(dataset_path)
print(f"Train: {len(train_x)} - {len(train_y)}")
print(f"Test: {len(test_x)} - {len(test_y)}")
train_dataset = tf_dataset(train_x, train_y, batch=batch_size)
test_dataset = tf_dataset(test_x, test_y, batch=batch_size)
""" Model """
model = build_unet(input_shape)
model.compile(loss="binary_crossentropy",
optimizer=tf.keras.optimizers.Adam(lr),
metrics=[
tf.keras.metrics.MeanIoU(num_classes=2),
tf.keras.metrics.Recall(),
tf.keras.metrics.Precision()
])
# model.summary()
callbacks = [
ModelCheckpoint(model_path, monitor="val_loss", verbose=1),
batch_size = args['batch_size']
output_dir = args['output_dir']
ROOT_DIR = os.path.abspath(os.curdir)
OUTPUT_DIR = os.path.join(ROOT_DIR, output_dir,
datetime.now().strftime("%d-%m-%YT%H-%M-%S"))
PREDICTIONS_DIR = os.path.join(OUTPUT_DIR, 'predictions')
if not os.path.exists(OUTPUT_DIR):
os.makedirs(OUTPUT_DIR)
if not os.path.exists(PREDICTIONS_DIR):
os.makedirs(PREDICTIONS_DIR)
(train_x, train_y), (valid_x, valid_y), (test_x, test_y) = load_data()
train_dataset = tf_dataset(train_x, train_y, batch=batch_size)
valid_dataset = tf_dataset(valid_x, valid_y, batch=batch_size)
test_dataset = tf_dataset(test_x, test_y, batch=batch_size)
with strategy.scope():
model = build_model()
opt = tf.keras.optimizers.Adam(learning_rate)
metrics = [
'acc',
tf.keras.metrics.Recall(),
tf.keras.metrics.Precision(), iou, dice_coefficient
]
model.compile(loss="binary_crossentropy",
optimizer=opt,
metrics=metrics)
