def check_transform(img_path, out_path):
imgset = imgs.make_imgs(img_path, norm=False, transform=imgs.to_3D)
dirs.make_dir(out_path)
for i, img_i in enumerate(imgset):
raw_i = img_i[0] # imgs.unify_img(img_i)
print(raw_i.shape)
name_i = "img" + str(i) + ".jpg"
new_img_i = imgs.Image(name_i, raw_i)
new_img_i.save(out_path)
def reconstruction(ae_path, img_path, out_path):
reader = deep.reader.NNReader()
nn = reader.read(ae_path) # autoconv.read_conv_ae(ae_path)
imgset = imgs.make_imgs(img_path, norm=True)
recon = [nn.reconstructed(img_i) for img_i in imgset]
recon = imgs.unorm(recon)
dirs.make_dir(out_path)
for i, img_i in enumerate(recon):
img_i.save(out_path, i)
def agum_data(action_path, out_path):
old_actions = actions.read_actions(action_path)
new_actions = [flip_action(action_i) for action_i in old_actions]
all_actions = old_actions + new_actions
dirs.make_dir(out_path)
for action_i in all_actions:
path_i = utils.paths.Path(out_path)
path_i.add(action_i.cat)
dirs.make_dir(path_i)
action_i.save(str(path_i))
def select_frames(action_path,out_path,action_dict):
print(action_dict.name)
action_path=action_path.create(action_dict.name)
out_path=out_path.create(action_dict.name)
for in_path_i,cat_i in action_dict.categorize(action_path):
#dst_path=out_path.create(action_dict.name)
dirs.make_dir(out_path)
dst_path_i=out_path.replace(in_path_i)
print(str(in_path_i))
print(str(dst_path_i))
copyfile(str(in_path_i), str(dst_path_i))
def save_actions(actions,outpath):
dirs.make_dir(outpath)
print(dir(utils.data))
extr_cats=utils.data.ExtractCat(parse_cat=lambda a:a.cat)
for action_i in actions:
extr_cats(action_i)
for name_i in extr_cats.names():
cat_dir_i=outpath.append(name_i,copy=True)
dirs.make_dir(cat_dir_i)
for action_i in actions:
cat_path_i=outpath.append(action_i.cat,copy=True)
action_i.save(cat_path_i)
def seg_action(action_path,out_path,action_dict):
action_path=action_path.create(action_dict.name)
out_path=out_path.create(action_dict.name)
print(str(action_path))
print(str(out_path))
cats=action_dict.categorize(action_path)
cat_dict={}
for cat_i in action_dict.data.keys():
cat_path=out_path.copy()
cat_path.set_name(cat_i)
dirs.make_dir(cat_path)
cat_dict[cat_i]=cat_path
for frame_path,cat_i in cats:
dst_path=cat_dict[cat_i]
dst_path=dst_path.create(action_dict.name)
dirs.make_dir(dst_path)
dst_path=dst_path.replace(frame_path)
print(frame_path)
print(dst_path)
copyfile(str(frame_path), str(dst_path))
import os
from utils.load_conf import parse_args, parse_conf
from utils.dirs import make_dir, clear_dirs
from run_main import run_timit
if __name__ == '__main__':
# Locate the root directory of the project
root = os.environ['VOICE_VAE']
# parse command line arguments
args = parse_args()
if args is None:
exit()
# parse configuration file arguments
config = parse_conf(root, args)
# Setup the output directories of the program
if config['clear']:
clear_dirs()
for n in ['results_path', 'results_dir', 'models_path', 'models_dir', 'tensorboard_dir', 'train_tb_dir', 'test_tb_dir']:
d = config[n]
make_dir(d)
# main
run_timit(config)
def extract_features(in_path,ext_path,seq_path='seq',short_names=True):
out_path=in_path.copy().set_name(str(seq_path))
dirs.make_dir(str(out_path))
extractor=ext.read_external(str(ext_path),short_names)
transform_seq(in_path,out_path,extractor)
def save(self,outpath):
full_outpath=outpath.append(self.name,copy=True)
dirs.make_dir(full_outpath)
[img_i.save(full_outpath)
for img_i in self.img_seq]
def extract(action_path,out_path,cat_path):
dirs.make_dir(out_path)
actions=parse_info(cat_path)
print(str(actions[0]))
[select_frames(action_path,out_path,act_i) for act_i in actions]
