def __init__(self, zFile):
output_dir = get_meta_dir()
assert output_dir.is_dir()
zTime = datetime.now().strftime("%Y%m%d-%H%M%S")
self._log_path = output_dir / (zFile + "_" + zTime + ".txt")
def __init__(self):
df_c = pd.read_feather(get_meta_dir() / "face_clusters.feather")
l_parts = []
for x in range(50):
path = get_part_dir(x, False)
isDir = path.is_dir()
if isDir:
l_parts.append(x)
l_orig = []
l_file = []
l_part = []
for iPart in l_parts:
df_meta = read_metadata(iPart)
for x in df_meta:
num_fakes = len (x[1])
l_orig.extend([x[0]]* (num_fakes + 1))
l_file.append(x[0])
l_file.extend(x[1])
l_part.extend([iPart] * (num_fakes + 1))
df = pd.DataFrame({'orig': l_orig, 'file': l_file, 'part': l_part})
df = df.merge(df_c, left_on = 'orig', right_on='video')
df = df.drop(['video', 'chunk'], axis = 1)
l_file_tuple = list(zip (df.file, df.part))
l_exists = []
for x in l_file_tuple:
filepath = get_part_dir(x[1]) / x[0]
l_exists.append(filepath.is_file())
df = df.assign(exists = l_exists)
num_files = df.shape[0]
num_originals = np.unique(df.orig).shape[0]
num_clusters = np.unique(df.cluster).shape[0]
# print(f"num_files = {num_files}, num_originals = {num_originals}, num_clusters = {num_clusters}")
self._df = df
num_originals = azOriginal.shape[0]
num_valid = int(1 + (rValidationSplit * num_originals))
num_train = num_originals - num_valid
azTest = azOriginal[:num_valid]
azTrain = azOriginal[num_valid:]
m_train = df.original.isin(azTrain)
m_test = df.original.isin(azTest)
assert (m_train ^ m_test).all()
df = df.assign(m_train=m_train, m_test=m_test)
df.to_pickle(get_meta_dir() / "df_tgs.pkl")
idx_train = np.where(m_train)[0]
# Todo: seed
np.random.shuffle(idx_train)
num_max_files_per_run = 7000
num_splits = int(1 + idx_train.shape[0] / num_max_files_per_run)
l_idx_train = np.array_split(idx_train, num_splits)
z_model_name = "my_keras"
checkpoint_path = str(get_model_dir() / f"{z_model_name}.model")
c("qrbjzz")
c("lssyhe")
c("qhqtm")
c("jqgia")
c("xjabn")
c("hpeya")
c("ggoq")
c("rfwek")
c("hmfc")
c("wervs")
c("dhwgib")
c("rzpsy")
input_dir = get_meta_dir()
df_all = pd.read_pickle(input_dir / "all_files.pkl")
l_original = []
l_part = []
def c(txt):
m = df_all.original.str.startswith(txt)
nOriginals = df_all.original[m].unique().shape[0]
assert nOriginals == 1
l_original.append(df_all.original[m].iloc[0])
img_size_target = 128
img_size_ori = 128
def upsample(img):
if img_size_ori == img_size_target:
return img
return resize(img, (img_size_target, img_size_target), mode='constant', preserve_range=True)
#res = np.zeros((img_size_target, img_size_target), dtype=img.dtype)
#res[:img_size_ori, :img_size_ori] = img
#return res
model = load_model(get_meta_dir() / "model_2", custom_objects = {'bce_dice_loss' : bce_dice_loss, 'my_iou_metric': my_iou_metric})
df = pd.read_pickle(get_meta_dir() / "df_tgs.pkl")
df = df[df.m_test]
m_fake = (df.original != df.file_stem)
df = df.assign(fake = m_fake)
g = df.groupby('file_stem')
s_files = g.file.apply(list)
s_masks = g.file_mask.apply(list)
s_target = g.fake.first()