# tensorboard 出力
board_train = tf.summary.create_file_writer(
logdir=os.path.join(args.tensorboard_dir, args.exper_name))
board_valid = tf.summary.create_file_writer(
logdir=os.path.join(args.tensorboard_dir, args.exper_name + "_valid"))
board_train.set_as_default()
#board_valid.set_as_default()
#================================
# データセットの読み込み
#================================
# 学習用データセットとテスト用データセットの設定
ds_train, ds_valid, n_trains, n_valids = load_dataset(
args.dataset_dir,
image_height=args.image_height,
image_width=args.image_width,
n_channels=3,
batch_size=args.batch_size,
use_tfrecord=args.use_tfrecord,
seed=args.seed)
if (args.debug):
print("n_trains : ", n_trains)
print("n_valids : ", n_valids)
print("ds_train : ", ds_train)
print("ds_valid : ", ds_valid)
#================================
# モデルの構造を定義する。
#================================
model_G = TempleteNetworks(out_dim=3)
if (args.debug):
model_G(
# データセットの読み込み
#================================
# 学習用データセットとテスト用データセットの設定
if (args.use_datagen):
datagen_train = TempleteDataGen(
dataset_dir=args.dataset_dir,
datamode="train",
image_height=args.image_height,
image_width=args.image_width,
batch_size=args.batch_size,
)
else:
image_s_trains, image_t_trains, image_s_valids, image_t_valids = load_dataset(
args.dataset_dir,
image_height=args.image_height,
image_width=args.image_width,
n_channels=3,
batch_size=args.batch_size,
seed=args.seed)
if (args.debug):
print("[image_s_trains] shape={}, dtype={}, min={}, max={}".format(
image_s_trains.shape, image_s_trains.dtype,
np.min(image_s_trains), np.max(image_s_trains)))
print("[image_t_trains] shape={}, dtype={}, min={}, max={}".format(
image_t_trains.shape, image_t_trains.dtype,
np.min(image_t_trains), np.max(image_t_trains)))
print("[image_s_valids] shape={}, dtype={}, min={}, max={}".format(
image_s_valids.shape, image_s_valids.dtype,
np.min(image_s_valids), np.max(image_s_valids)))
print("[image_t_valids] shape={}, dtype={}, min={}, max={}".format(
image_t_valids.shape, image_t_valids.dtype,
# fetch model
if args.model == 'dae':
agent = DAE(params, logger, args.gpu)
elif args.model == 'protonet':
agent = SupportQueryAgent(params, logger, args.gpu)
elif args.model == 'dve':
agent = DVE(params, logger, args.gpu)
elif args.model == 'daie':
agent = DAIE(params, logger, args.gpu)
else:
raise NotImplementedError
# get dataset
dataset = load_dataset(params)
valid, test = None, None
if len(params['data']['split']) == 1:
train = dataset['train']
elif len(params['data']['split']) == 2:
train, test = dataset['train'], dataset['test']
elif len(params['data']['split']) == 3:
train, valid, test = dataset['train'], dataset['valid'], dataset['test']
# start training
if args.type == 'train':
if args.pretrain_dir is not None:
agent.start(train, args.pretrain_dir)
else: