def maybe_download_and_extract():
"""
Download and extract the weather-data if the data-files don't
already exist in the data_dir.
"""
download.maybe_download_and_extract(url=data_url, download_dir=data_dir)
def maybe_download():
print("Downloading Inception 5h Model ...")
# The file on the internet is not stored in a compressed format.
# This function should not extract the file when it does not have
# a relevant filename-extensions such as .zip or .tar.gz
download.maybe_download_and_extract(url=data_url, download_dir=data_dir)
def maybe_download_and_extract():
"""
data_path(원하는 경로를 처음에 설정해라) 에 존재하지 않는다면 CIFAR-10 데이터셋을
다운로드 하고 추출한다
"""
download.maybe_download_and_extract(url=data_url, download_dir=data_path)
def maybe_download_and_extract():
"""
Download and extract the CIFAR-10 data-set if it doesn't already exist
in data_path (set this variable first to the desired path).
"""
download.maybe_download_and_extract(url=data_url, download_dir=data_path)
def maybe_download_and_extract():
"""
Download and extract the data-set if it doesn't already exist
in data_dir (set this variable first to the desired directory).
"""
download.maybe_download_and_extract(url=data_url, download_dir=data_dir)
def maybe_download_and_extract():
"""
Download and extract the weather-data if the data-files don't
already exist in the data_dir.
"""
download.maybe_download_and_extract(url=data_url, download_dir=data_dir)
def download_data(in_dir, url):
if not os.path.exists(in_dir):
os.makedirs(in_dir)
download.maybe_download_and_extract(url,in_dir)
def maybe_download_and_extract():
"""
Download and extract the IMDB Review data-set if it doesn't already exist
in data_dir (set this variable first to the desired directory).
"""
download.maybe_download_and_extract(url=data_url, download_dir=data_dir)
def maybe_download_and_extract():
"""
Download and extract the CIFAR-10 data-set if it doesn't already exist
in data_path (set this variable first to the desired path).
"""
download.maybe_download_and_extract(url=data_url, download_dir=data_path)
def download_data(in_dir, url):
# Si la carpeta no existe la creamos
if not os.path.exists(in_dir):
os.makedirs(in_dir)
# Para descargar del link directo y extraer los archivos
download.maybe_download_and_extract(url, in_dir)
def maybe_download():
"""
Download the Inception model from the internet if it does not already
exist in the data_dir. The file is about 50 MB.
"""
print("Downloading Inception 5h Model ...")
download.maybe_download_and_extract(url=data_url, download_dir=data_dir)
def maybe_download():
"""
Download the Inception model from the internet if it does not already
exist in the data_dir. The file is about 85 MB.
"""
print("Downloading Inception v3 Model ...")
download.maybe_download_and_extract(url=data_url, download_dir=data_dir)
def maybe_download():
"""
data_dir에 존재하지 않으면, 인터넷으로부터 인셉션 모델을 다운로드한다
이 파일은 약 85 MB 다.
"""
print("Downloading Inception v3 Model ...")
download.maybe_download_and_extract(url=data_url, download_dir=data_dir)
def maybe_download_and_extract(data_path):
# URL for the data-set on the internet.
data_url = "https://www.cs.toronto.edu/~kriz/cifar-10-python.tar.gz"
"""
Download and extract the CIFAR-10 data-set if it doesn't already exist
in data_path (set this variable first to the desired path).
"""
download.maybe_download_and_extract(url=data_url, download_dir=data_path)
def maybe_download():
# Download the VGG16 model from the internet if it does not already
# exist in the data_dir. WARNING! The file is about 550 MB.
print("Downloading VGG16 Model ...")
# The file on the internet is not stored in a compressed format.
# This function should not extract the file when it does not have
# a relevant filename-extensions such as .zip or .tar.gz
download.maybe_download_and_extract(url=data_url, download_dir=data_dir)
def maybe_download_and_extract():
filenames = [
"zips/train2017.zip", "zips/val2017.zip",
"annotations/annotations_trainval2017.zip"
]
for filename in filenames:
url = data_url + filename
print("Downloading " + url)
download.maybe_download_and_extract(url=url, download_dir=data_dir)
def maybe_download():
"""
data_dir 안에 이미 존재하지 않으면, 인터넷으로부터 VGG16 모델을
다운로드 받는다. 경고! 이 파일은 약 550 MB 이다.
"""
print("Downloading VGG16 Model ...")
# 인터넷에서 이 파일은 압축된 포맷으로 저장되어 있지 않다.
# 이 함수는 파일 확장자가 .zip 이나 tar.gz 이 아닌 경우에 추출해서는 안 된다.
download.maybe_download_and_extract(url=data_url, download_dir=data_dir)
def maybe_download():
"""
Baixa o modelo VGG16 da internet se ainda não
existe no data_dir. ATENÇÃO! O arquivo tem cerca de 550 MB.
"""
print("Downloading VGG16 Model ...")
# O arquivo na internet não é armazenado em um formato comprimido.
# Esta função não deve extrair o arquivo quando não tiver
# um arquivo relevante - extensões como .zip ou .tar.gz
download.maybe_download_and_extract(url=data_url, download_dir=data_dir)
def maybe_download_and_extract(language_code="da"):
"""
Download and extract the Europarl data-set if the data-file doesn't
already exist in data_dir. The data-set is for translating between
English and the given language-code (e.g. 'da' for Danish, see the
list of available language-codes above).
"""
# Create the full URL for the file with this data-set.
url = data_url + language_code + "-en.tgz"
download.maybe_download_and_extract(url=url, download_dir=data_dir)
def maybe_download_and_extract(language_code="da"):
"""
Download and extract the Europarl data-set if the data-file doesn't
already exist in data_dir. The data-set is for translating between
English and the given language-code (e.g. 'da' for Danish, see the
list of available language-codes above).
"""
# Create the full URL for the file with this data-set.
url = data_url + language_code + "-en.tgz"
download.maybe_download_and_extract(url=url, download_dir=data_dir)
def maybe_download():
"""
Download the VGG16 model from the internet if it does not already
exist in the data_dir. WARNING! The file is about 550 MB.
"""
print("Downloading VGG16 Model ...")
# The file on the internet is not stored in a compressed format.
# This function should not extract the file when it does not have
# a relevant filename-extensions such as .zip or .tar.gz
download.maybe_download_and_extract(url=data_url, download_dir=data_dir)
def downloader(url, path):
download.maybe_download_and_extract(url, path)
dirCreator(meta, path)
for batch in batches:
print(batch)
sep = batchLoader(batch)
if batch == validation:
sepSaver(sep, 'validation', batch)
elif batch == test:
sepSaver(sep, 'test', batch)
else:
sepSaver(sep, 'train', batch)
def maybe_download_and_extract():
"""
Download and extract the COCO data-set if the data-files don't
already exist in data_dir.
"""
# Filenames to download from the internet.
filenames = ["zips/train2017.zip", "zips/val2017.zip",
"annotations/annotations_trainval2017.zip"]
# Download these files.
for filename in filenames:
# Create the full URL for the given file.
url = data_url + filename
print("Downloading " + url)
download.maybe_download_and_extract(url=url, download_dir=data_dir)
acc_list = []
count = 0
retrain = 0
parent_dir = 'assets/'
# lr = 1e-5
lr = 1e-4
crates = {'cov1': 0., 'cov2': 0., 'fc1': 0., 'fc2': 0., 'fc3': 0.}
retrain_cnt = 0
roundrobin = 0
with_biases = False
# check npy files
url = 'http://www.cs.toronto.edu/~guerzhoy/tf_alexnet/bvlc_alexnet.npy'
download_dir = '.'
download.maybe_download_and_extract(url, download_dir)
INITIAL_TRAIN = True
if (INITIAL_TRAIN):
# TRAIN
param = [
('-cRates', crates),
('-first_time', True),
# ('-first_time', False),
('-train', False),
('-test', True),
('-prune', False),
# ('-test', True),
('-lr', lr),
('-with_biases', with_biases),
('-parent_dir', parent_dir),
('-lambda1', 1e-5),
def main(data_set, data_dir='data/', create_csv_lists=True):
# urls for all data sets:
url_iiw = 'http://labelmaterial.s3.amazonaws.com/release/iiw-dataset-release-0.zip'
url_sintel_complete = 'http://files.is.tue.mpg.de/sintel/MPI-Sintel-complete.zip'
url_sintel_albedo = 'http://files.is.tue.mpg.de/jwulff/sintel/extras/MPI-Sintel-training_albedo_noshadingtextures.zip'
url_sintel_images = 'http://files.is.tue.mpg.de/jwulff/sintel/extras/MPI-Sintel-training_clean_noshadingtextures.zip'
url_sintel_shading = 'http://files.is.tue.mpg.de/jwulff/sintel/extras/MPI-Sintel-training_shading.zip'
url_mit = 'http://people.csail.mit.edu/rgrosse/intrinsic/intrinsic-data.tar.gz'
# directory of data
data_dir_sintel_complete = data_dir + 'mpi-sintel-complete/'
data_dir_sintel_shading = data_dir + 'mpi-sintel-shading/'
data_dir_mit = data_dir + 'MIT-intrinsic/data/'
data_dir_iiw = data_dir + 'iiw-dataset/data/'
if data_set == 'iiw':
download.maybe_download_and_extract(url=url_iiw, download_dir=data_dir)
if create_csv_lists:
# import file names of data directory:
df_iiw = pd.DataFrame(
[[
int(os.path.splitext(os.path.basename(x))[0]),
os.path.relpath(x, data_dir),
os.path.splitext(os.path.relpath(x, data_dir))[0] + '.json'
] for x in glob.glob(data_dir_iiw + '/*.png')],
columns=['file_id', 'image_path', 'label_path'])
# sort by file ids (we can sort these files because they are
# shuffled during training in tf anyways):
df_iiw.sort_values(by='file_id', inplace=True)
# reset indices of pd.DataFrame:
df_iiw.reset_index(drop=True, inplace=True)
# get training validation and testing data set of the iiw data:
df_iiw_train, df_iiw_valid, df_iiw_test, df_iiw_train_sample, \
df_iiw_valid_sample, \
df_iiw_test_sample = create_datasets_iiw(df=df_iiw, p_train=0.8,
data_dir=data_dir,
p_valid=0.1,
p_test=0.1,
sample=True)
# save complete data set, training data set, validation data set and
# testing data set in separate data files:
df_iiw.to_csv(path_or_buf=data_dir + 'data_iiw_complete.csv',
sep=',',
columns=['image_path', 'label_path'],
index=False,
header=False)
df_iiw_train.to_csv(path_or_buf=data_dir + 'data_iiw_train.csv',
sep=',',
columns=['image_path', 'label_path'],
index=False,
header=False)
df_iiw_valid.to_csv(path_or_buf=data_dir + 'data_iiw_valid.csv',
sep=',',
columns=['image_path', 'label_path'],
index=False,
header=False)
df_iiw_test.to_csv(path_or_buf=data_dir + 'data_iiw_test.csv',
sep=',',
columns=['image_path', 'label_path'],
index=False,
header=False)
df_iiw_train_sample.to_csv(path_or_buf=data_dir +
'sample_data_iiw_train.csv',
sep=',',
columns=['image_path', 'label_path'],
index=False,
header=False)
df_iiw_valid_sample.to_csv(path_or_buf=data_dir +
'sample_data_iiw_valid.csv',
sep=',',
columns=['image_path', 'label_path'],
index=False,
header=False)
df_iiw_test_sample.to_csv(path_or_buf=data_dir +
'sample_data_iiw_test.csv',
sep=',',
columns=['image_path', 'label_path'],
index=False,
header=False)
elif data_set == 'mpi_sintel_complete':
data_dir_sintel_complete = data_dir + 'mpi-sintel-complete/'
download.maybe_download_and_extract(
url=url_sintel_complete, download_dir=data_dir_sintel_complete)
if create_csv_lists:
# use 'clean pass' images (see narihira2015: p.3: "'final images'
# [...] are the result of additional computer graphics tricks which
# dristract from our application."):
df_sintel = pd.DataFrame(
[[
os.path.relpath(x, data_dir),
os.path.relpath(x, data_dir).replace('clean', 'albedo'),
os.path.relpath(x, data_dir).replace('clean', 'invalid')
] for x in glob.glob(data_dir_sintel_complete +
'training/clean/**/*.png')],
columns=['image_path', 'label_path', 'invalid_path'])
# add scene to dataframe (for splitting into train/valid/test sets):
df_sintel['scene_c'] = df_sintel['image_path'].apply(
lambda row: row.split('/')[-2:-1]).apply(pd.Series)
# get training validation and testing data set of the mpi-sintel
# data:
df_sintel_train, df_sintel_valid, df_sintel_test, df_sintel_train_sample, \
df_sintel_valid_sample, df_sintel_test_sample = create_datasets_sintel(df_sintel)
# save complete data set, training data set, validation data set and
# testing data set in separate data files:
df_sintel.to_csv(
path_or_buf=data_dir + 'data_sintel_complete_complete.csv',
sep=',',
columns=['image_path', 'label_path', 'invalid_path'],
index=False,
header=False)
df_sintel_train.to_csv(
path_or_buf=data_dir + 'data_sintel_complete_train.csv',
sep=',',
columns=['image_path', 'label_path', 'invalid_path'],
index=False,
header=False)
df_sintel_valid.to_csv(
path_or_buf=data_dir + 'data_sintel_complete_valid.csv',
sep=',',
columns=['image_path', 'label_path', 'invalid_path'],
index=False,
header=False)
df_sintel_test.to_csv(
path_or_buf=data_dir + 'data_sintel_complete_test.csv',
sep=',',
columns=['image_path', 'label_path', 'invalid_path'],
index=False,
header=False)
df_sintel_train_sample.to_csv(
path_or_buf=data_dir + 'sample_data_sintel_complete_train.csv',
sep=',',
columns=['image_path', 'label_path', 'invalid_path'],
index=False,
header=False)
df_sintel_valid_sample.to_csv(
path_or_buf=data_dir + 'sample_data_sintel_complete_valid.csv',
sep=',',
columns=['image_path', 'label_path', 'invalid_path'],
index=False,
header=False)
df_sintel_test_sample.to_csv(
path_or_buf=data_dir + 'sample_data_sintel_complete_test.csv',
sep=',',
columns=['image_path', 'label_path', 'invalid_path'],
index=False,
header=False)
# also save (unknown) test files:
df_sintel_test_unknown = pd.DataFrame(
[[os.path.relpath(x, data_dir), None, None]
for x in glob.glob(data_dir_sintel_complete +
'test/clean/**/*.png')],
columns=['image_path', 'label_path', 'invalid_path'])
df_sintel_test_unknown.to_csv(
path_or_buf=data_dir + 'data_sintel_complete_test_unknown.csv',
sep=',',
columns=['image_path', 'label_path', 'invalid_path'],
index=False,
header=False)
elif data_set == 'mpi_sintel_shading':
# The problem is that the shading files (\*\*/out\_*.png)
# are named differently than the clean/albodo files
# (\*\*/frame\_\*.png).
# Also their numbering does not start with 1, 2, ...
# Therefore we import each file path (clean, albedo and shading)
# separately, sort it by the scene and frame (increasing) and
# merge the three paths. Furthermore we have to get rid of some
# scenes which are not included in either clean or albedo or shading.
# maybe download data if necessary:
download.maybe_download_and_extract(url=url_sintel_images,
download_dir=data_dir +
'mpi-sintel-shading/')
download.maybe_download_and_extract(url=url_sintel_albedo,
download_dir=data_dir +
'mpi-sintel-shading/')
download.maybe_download_and_extract(url=url_sintel_shading,
download_dir=data_dir +
'mpi-sintel-shading/')
if create_csv_lists:
# import images and labels separateley:
df_clean = pd.DataFrame([
os.path.relpath(x, data_dir)
for x in glob.glob(data_dir_sintel_shading +
'clean_noshadingtextures/**/*.png')
],
columns=['image_path'])
df_clean[['scene', 'frame']] = df_clean['image_path'].apply(
lambda row: row.split('/')[-2:]).apply(pd.Series)
df_albedo = pd.DataFrame([
os.path.relpath(x, data_dir)
for x in glob.glob(data_dir_sintel_shading +
'albedo_noshadingtextures/**/*.png')
],
columns=['albedo_label_path'])
df_albedo[['scene',
'frame']] = df_albedo['albedo_label_path'].apply(
lambda row: row.split('/')[-2:]).apply(pd.Series)
df_shading = pd.DataFrame([
os.path.relpath(x, data_dir)
for x in glob.glob(data_dir_sintel_shading +
'shading/**/*.png')
],
columns=['shading_label_path'])
df_shading[['scene',
'frame']] = df_shading['shading_label_path'].apply(
lambda row: row.split('/')[-2:]).apply(pd.Series)
try:
df_invalid = pd.DataFrame([
os.path.relpath(x, data_dir)
for x in glob.glob(data_dir_sintel_complete +
'training/invalid/**/*.png')
],
columns=['invalid_path'])
df_invalid[['scene',
'frame']] = df_invalid['invalid_path'].apply(
lambda row: row.split('/')[-2:]).apply(
pd.Series)
except KeyError:
print('We need to download and extract the ' +
'mpi_sintel_complete dataset first to get the invalid ' +
'pixel mask.')
main(data_set='mpi_sintel_complete',
data_dir=data_dir,
create_csv_lists=False)
df_invalid = pd.DataFrame([
os.path.relpath(x, data_dir)
for x in glob.glob(data_dir_sintel_complete +
'training/invalid/**/*.png')
],
columns=['invalid_path'])
df_invalid[['scene',
'frame']] = df_invalid['invalid_path'].apply(
lambda row: row.split('/')[-2:]).apply(
pd.Series)
# get list which contains scenes which have to be deleted:
lst_del = [
list(df_albedo[~df_albedo['scene'].isin(
df_clean['scene'].unique())]['scene'].unique()) +
list(df_clean[~df_clean['scene'].isin(
df_shading['scene'].unique())]['scene'].unique()) +
list(df_shading[~df_shading['scene'].isin(
df_invalid['scene'].unique())]['scene'].unique()) +
list(df_invalid[~df_invalid['scene'].isin(
df_albedo['scene'].unique())]['scene'].unique())
][0]
# delete scenes from lst_del, sort by ('scene', 'frame') and reset
# index:
df_clean = df_clean[~df_clean['scene'].isin(lst_del)]
df_clean = df_clean.sort_values(['scene', 'frame'])
df_clean.reset_index(drop=True, inplace=True)
df_albedo = df_albedo[~df_albedo['scene'].isin(lst_del)]
df_albedo = df_albedo.sort_values(['scene', 'frame'])
df_albedo.reset_index(drop=True, inplace=True)
df_shading = df_shading[~df_shading['scene'].isin(lst_del)]
df_shading = df_shading.sort_values(['scene', 'frame'])
df_shading.reset_index(drop=True, inplace=True)
df_invalid = df_invalid[~df_invalid['scene'].isin(lst_del)]
df_invalid = df_invalid.sort_values(['scene', 'frame'])
df_invalid.reset_index(drop=True, inplace=True)
# merge all four DataFrames and keep just important paths:
df_merged = df_clean.merge(df_albedo,
left_index=True,
right_index=True,
how='inner',
suffixes=('_c', '_a'))
df_merged = df_merged.merge(df_shading,
left_index=True,
right_index=True,
how='inner',
suffixes=('', '_s'))
df_sintel2 = df_merged.merge(df_invalid,
left_index=True,
right_index=True,
how='inner',
suffixes=('_s', '_i'))[[
'image_path', 'albedo_label_path',
'shading_label_path',
'invalid_path', 'scene_c'
]]
df_sintel_train2, df_sintel_valid2, df_sintel_test2, df_sintel_train_sample2, \
df_sintel_valid_sample2, df_sintel_test_sample2 = create_datasets_sintel(df_sintel2)
# save complete data set, training data set, validation data set and
# testing data set in separate data files:
df_sintel2.to_csv(
path_or_buf=data_dir + 'data_sintel_shading_complete.csv',
sep=',',
columns=[
'image_path', 'albedo_label_path', 'shading_label_path',
'invalid_path'
],
index=False,
header=False)
df_sintel_train2.to_csv(
path_or_buf=data_dir + 'data_sintel_shading_train.csv',
sep=',',
columns=[
'image_path', 'albedo_label_path', 'shading_label_path',
'invalid_path'
],
index=False,
header=False)
df_sintel_valid2.to_csv(
path_or_buf=data_dir + 'data_sintel_shading_valid.csv',
sep=',',
columns=[
'image_path', 'albedo_label_path', 'shading_label_path',
'invalid_path'
],
index=False,
header=False)
df_sintel_test2.to_csv(
path_or_buf=data_dir + 'data_sintel_shading_test.csv',
sep=',',
columns=[
'image_path', 'albedo_label_path', 'shading_label_path',
'invalid_path'
],
index=False,
header=False)
df_sintel_train_sample2.to_csv(
path_or_buf=data_dir + 'sample_data_sintel_shading_train.csv',
sep=',',
columns=[
'image_path', 'albedo_label_path', 'shading_label_path',
'invalid_path'
],
index=False,
header=False)
df_sintel_valid_sample2.to_csv(
path_or_buf=data_dir + 'sample_data_sintel_shading_valid.csv',
sep=',',
columns=[
'image_path', 'albedo_label_path', 'shading_label_path',
'invalid_path'
],
index=False,
header=False)
df_sintel_test_sample2.to_csv(
path_or_buf=data_dir + 'sample_data_sintel_shading_test.csv',
sep=',',
columns=[
'image_path', 'albedo_label_path', 'shading_label_path',
'invalid_path'
],
index=False,
header=False)
elif data_set == 'mit':
# maybe download data if necessary:
download.maybe_download_and_extract(url=url_mit, download_dir=data_dir)
if create_csv_lists:
df_mit = pd.DataFrame([[
os.path.relpath(x, data_dir),
os.path.relpath(x, data_dir).replace('original',
'reflectance'),
os.path.relpath(x, data_dir).replace('original', 'shading')
] for x in glob.glob(data_dir_mit + '**/original.png')],
columns=[
'image_path', 'albedo_label_path',
'shading_label_path'
])
# get training validation and testing data set of the mit data:
df_mit_train, df_mit_valid, df_mit_test, _, _, \
_ = create_datasets_mit(df=df_mit, p_train=0.8,
p_valid=0.1, p_test=0.1, sample=False)
# save complete data set, training data set, validation data set and
# testing data set in separate data files:
df_mit.to_csv(path_or_buf=data_dir + 'data_mit_complete.csv',
sep=',',
columns=[
'image_path', 'albedo_label_path',
'shading_label_path'
],
index=False,
header=False)
df_mit_train.to_csv(path_or_buf=data_dir + 'data_mit_train.csv',
sep=',',
columns=[
'image_path', 'albedo_label_path',
'shading_label_path'
],
index=False,
header=False)
df_mit_valid.to_csv(path_or_buf=data_dir + 'data_mit_valid.csv',
sep=',',
columns=[
'image_path', 'albedo_label_path',
'shading_label_path'
],
index=False,
header=False)
df_mit_test.to_csv(path_or_buf=data_dir + 'data_mit_test.csv',
sep=',',
columns=[
'image_path', 'albedo_label_path',
'shading_label_path'
],
index=False,
header=False)
else:
raise ValueError("data_set must be in ['iiw', 'mpi_sintel_shading', " +
"'mpi_sintel_complete', 'mit']")
def maybe_download_and_extract(language_code="el"):
# Create the full URL for the file with this data-set.
url = data_url + language_code + "-en.tgz"
download.maybe_download_and_extract(url=url, download_dir=data_dir)
def maybe_download():
print("Downloading Inception 5h Model ...")
download.maybe_download_and_extract(url=data_url, download_dir=data_dir)
def maybe_download_and_extract(language_code="da"):
url = data_url + language_code + "-en.tgz"
download.maybe_download_and_extract(url=url, download_dir=data_dir)
def maybe_download():
print("Downloading VGG16 Model ...")
download.maybe_download_and_extract(url=data_url, download_dir=data_dir)
def get_mnist_data(url, data_dir):
print("Downloading {} into {}".format(url, data_dir))
download.maybe_download_and_extract(url, data_dir)
def maybe_download_and_extract():
url = 'https://www.cs.toronto.edu/~kriz/cifar-10-python.tar.gz'
download.maybe_download_and_extract(url, '.')
def download_data():
download.maybe_download_and_extract(url=data_url, download_dir=data_path)
download.maybe_download(url=mat_url, download_dir=data_path)
# https://s3.amazonaws.com/cadl/models/synset.txt
# Internet URL for the file with the VGG16 model.
# Note that this might change in the future and will need to be updated.
data_url = "https://s3.amazonaws.com/cadl/models/vgg16.tfmodel"
# Directory to store the downloaded data.
data_dir = "vgg16/"
# File containing the TensorFlow graph definition. (Downloaded)
path_graph_def = "vgg16.tfmodel"
#########i###############################################################
if not os.path.exists(data_dir):
os.makedirs(data_dir)
"""
Download the VGG16 model from the internet if it does not already
exist in the data_dir. WARNING! The file is about 550 MB.
"""
print("Downloading VGG16 Model ...")
# The file on the internet is not stored in a compressed format.
# This function should not extract the file when it does not have
# a relevant filename-extensions such as .zip or .tar.gz
download.maybe_download_and_extract(url=data_url, download_dir=data_dir)
def maybe_download(self):
download.maybe_download_and_extract(url=self.data_url,
download_dir=self.data_dir)
def maybe_download_and_extract():
# Download and extract the Knifey-Spoony data-set if it doesn't already exit
# in data_dir (set this variable first to the desired directory).
download.maybe_download_and_extract(url=data_url, download_dir=data_dir)
def maybe_download_and_extract():
download.maybe_download_and_extract(url=data_url, download_dir=data_path)
