def get_data(self):
    def _parse_function(xfilename,yfilename):
      W=960
      H=720
      image_size=vgg.vgg_16.default_image_size
      image_string = tf.read_file(xfilename)
      image_decoded = tf.image.decode_jpeg(image_string, channels=3)          # (1)
      image = tf.cast(image_decoded, tf.float32)
      x=tf.image.resize(image,(image_size,image_size))

      image_string = tf.read_file(yfilename)
      image_decoded = tf.image.decode_jpeg(image_string, channels=3)          # (1)
      y = tf.cast(image_decoded, tf.int32)

      equality = tf.equal(tf.reshape(y,[W, H, 1, 3] ), tf.reshape(classes, [n_class,3]))
      equality = tf.cast(tf.reduce_all(equality, axis=-1), tf.int32)
      return x,equality

    dataset=tf.data.Dataset.from_tensor_slices((self.xfilenames,self.yfilenames))
    dataset=dataset.map(_parse_function,num_parallel_calls=8)
    dataset=dataset.batch(self.batch_size)
    self.dataset=dataset
    iterator = tf.data.Iterator.from_structure(self.dataset.output_types,
                                               self.dataset.output_shapes)
    self.img, self.label = iterator.get_next()
    print("get_data label",self.label.shape)
    self.train_init = iterator.make_initializer(self.dataset)    # initializer for train_data