def gen_data_label(file_name, is_train):
input_class = data_class.DataClass(tf.constant([], tf.string))
input_class.decode_class = data_class.BINClass(
[FLAGS.feature_row, FLAGS.feature_col, FLAGS.feature_cha])
label_class = data_class.DataClass(tf.constant([], tf.string))
label_class.decode_class = data_class.BINClass(
[FLAGS.label_row, FLAGS.label_col, FLAGS.label_cha])
tensor_list = [input_class] + [label_class]
file_queue = tensor_data.file_queue(file_name, is_train)
batch_tensor_list = tensor_data.file_queue_to_batch_data(
file_queue, tensor_list, is_train, FLAGS.batch_size)
return batch_tensor_list
def data_load(self, file_name):
is_train = True
st_data = data_class.DataClass(tf.constant([], tf.string))
st_data.decode_class = data_class.BINClass((self.st_len, 1))
image_data = data_class.DataClass(tf.constant([], tf.string))
image_data.decode_class = data_class.JPGClass(
(self.iheight, self.iwidth), 1, 0)
tensor_list = [st_data] + [image_data]
file_queue = tensor_data.file_queue(file_name, is_train)
batch_tensor_list = tensor_data.file_queue_to_batch_data(
file_queue, tensor_list, is_train, self.bsize, False)
self.st_data = batch_tensor_list[0]
self.image_data = batch_tensor_list[1]