def merge_files(src_directory: str,
output_shape: Tuple[int, int, int] = (1200, 1600, 3),
input_shape: Tuple[int, int] = (200, 200),
pocket: Tuple[int, int] = (100, 100),
output_filename: str = 'out.png'):
container = np.zeros(output_shape)
images = os.listdir(src_directory)
vertical_step, horizontal_step = input_shape
vertical_pocket, horizontal_pocket = pocket
with cb(f'Merging {src_directory}: ', max=len(images)) as bar:
for image in images:
row, column = parse_name(image)
start_row = row * (vertical_step - vertical_pocket)
start_column = column * (horizontal_step - horizontal_pocket)
end_row = start_row + vertical_step
end_column = start_column + horizontal_step
container[start_row:end_row, start_column:end_column] = np.array(
Image.open(os.path.join(src_directory, image)))
bar.next()
container = container.astype('uint8')
img = Image.fromarray(container)
img.save(output_filename)
def merge_directories(directories: list[str],
output_shape: Tuple[int, int, int] = (1200, 1600, 3),
input_shape: Tuple[int, int] = (200, 200),
pocket: Tuple[int, int] = (100, 100),
output_filename='out.png'):
with cb('Merging', max=len(directories)) as bar:
for directory in directories:
merge_files(directory, output_shape, input_shape, pocket,
f'{directory}_{output_filename}')
bar.next()
def split_images_in_directory(directory_name: str,
output_shape: Tuple[int, int] = (200, 200),
pocket: Tuple[int, int] = (100, 100),
extension: str = '.png'):
if not os.path.isdir(directory_name):
raise NotADirectoryError()
files = os.listdir(directory_name)
with cb('Splitting', max=len(files)) as bar:
for f in files:
file_path = os.path.join(directory_name, f)
if not os.path.isdir(file_path):
dir_name = f'out_{directory_name}'
if not os.path.exists(dir_name):
os.makedirs(dir_name)
split_image(file_path, dir_name, output_shape, pocket,
extension)
bar.next()
def make_predictions_for_images_in_directory(model: tf.keras.Model,
input_directory: str,
output_directory: str = 'out'):
if not os.path.exists(input_directory):
raise ValueError('Directory does not exists')
if not os.path.isdir(input_directory):
raise NotADirectoryError()
if not os.path.exists(output_directory):
os.makedirs(output_directory)
# Loading data
images = os.listdir(input_directory)
with cb('Predicting', max=len(images)) as b:
for image in images:
make_prediction_for_image(
model, os.path.join(input_directory, image),
os.path.join(output_directory,
f"{os.path.splitext(image)[0]}_prediction.png"))
b.next()
print('Finished!')
def blur_dir(dir, parameter=2):
files = os.listdir(dir)
with cb('Bluring', max=len(files)) as bar:
for file in files:
blur_file(os.path.join(dir, file), parameter)
bar.next()
def remove_alpha_dir(dir, threshold=125):
files = os.listdir(dir)
with cb('Removeing alpha', max=len(files)) as bar:
for file in files:
remove_alpha_file(os.path.join(dir, file), threshold)
bar.next()