mask_names.append(bbox_path)
#gt_images.append(image)
if i % batch_size == 0:
yield (gt_images[(i - batch_size):i],
input_images[(i - batch_size):i],
file_names[(i - batch_size):i],
mask_names[(i - batch_size):i])
#input_images = np.array(input_images)
#print('Shape of image: {}'.format(input_images.shape))
if __name__ == "__main__":
ng.get_gpus(1)
args = parser.parse_args()
model = InpaintGCModel()
sess_config = tf.ConfigProto()
sess_config.gpu_options.allow_growth = True
with tf.Session(config=sess_config) as sess:
input_image = tf.placeholder(tf.float32, shape=[1, 256, 256, 3])
input_mask = tf.placeholder(tf.float32, shape=[1, 256, 256, 1])
input_guide = tf.placeholder(tf.float32, shape=[1, 256, 256, 1])
output = model.build_server_graph(input_image, input_mask, input_guide)
output = (output + 1.) * 127.5
output = tf.reverse(output, [-1])
output = tf.saturate_cast(output, tf.uint8)
# load pretrained model
vars_list = tf.get_collection(tf.GraphKeys.GLOBAL_VARIABLES)
assign_ops = []
else:
data_mask_data = DataFromFNames(fnames,
config.IMG_SHAPES,
random_crop=config.RANDOM_CROP)
images = data_mask_data.data_pipeline(config.BATCH_SIZE)
masks = None
# # Mask Data
# with open(config.DATA_FLIST[config.MASKDATASET][0]) as f:
# fnames = f.read().splitlines()
# mask_data = MaskFromFNames(
# fnames, config.MASK_SHAPES, random_crop=config.RANDOM_CROP)
# masks = mask_data.data_pipeline(config.BATCH_SIZE)
guides = None
# main model
model = InpaintGCModel()
g_vars, d_vars, losses = model.build_graph_with_losses(images,
masks,
guides,
config=config)
# validation images
if config.VAL:
with open(config.DATA_FLIST[config.DATASET][1]) as f:
val_fnames = f.read().splitlines()
with open(config.DATA_FLIST[config.MASKDATASET][1]) as f:
val_mask_fnames = f.read().splitlines()
# progress monitor by visualizing static images
for i in range(config.STATIC_VIEW_SIZE):
static_fnames = val_fnames[i:i + 1]
if config.MASKFROMFILE: