def extract_batch(dataset, batch_size, is_training):
with tf.device("/cpu:0"):
data_provider = slim.dataset_data_provider.DatasetDataProvider(
dataset, num_readers=4, shuffle=False, common_queue_capacity=512, common_queue_min=32)
image, gt_mask = data_provider.get(['image', 'segmentation'])
image, gt_mask = preprocess.preprocess_image(image, gt_mask, is_training=is_training)
return tf.train.shuffle_batch([image, gt_mask], batch_size, 4096, 64, num_threads=4)
def read_data(is_training, split_name):
file_pattern = '{}_{}.tfrecord'.format(args.data_name, split_name)
tfrecord_path = os.path.join(args.data_dir,'records',file_pattern)
if is_training:
dataset = get_dataset(tfrecord_path)
image, gt_mask = extract_batch(dataset, args.batch_size, is_training)
else:
image, gt_mask = read_tfrecord(tfrecord_path)
image, gt_mask = preprocess.preprocess_image(image, gt_mask, is_training)
return image, gt_mask
