init_op = tf.initialize_all_variables()
with tf.Session() as sess:
  sess.run(init_op)
  saver.save(sess, "model.ckpt")

saver = tf.train.import_meta_graph("model.ckpt.meta")
with tf.Session() as sess:
  saver.restore(sess, "model.ckpt")

BOTTLENECK_TENSOR_NAME = 'pool_3/_reshape:0'
JPEG_DATA_TENSOR_NAME = 'DecodeJpeg/contents:0'
#读取已经训练好的模型
  with gfile.FastGFile(os.path.join(MODEL_DIR, MODEL_FILE), 'rb') as f:
    graph_def = tf.GraphDef()
    graph_def.ParseFromString(f.read())
  bottleneck_tensor, jpeg_data_tensor = tf.import_graph_def(graph_def, return_elements=[BOTTLENECK_TENSOR_NAME, JPEG_DATA_TENSOR_NAME])

def run_bottlenect_on_images(sess, image_data, image_data_tensor, bottlenect_tensor):
  bottlenect_values = sess.run(bottlenect_tensor, {image_data_tensor: image_data})

  # 经过卷积网络处理后的是一个思维数组，压缩成一个特征，一维向量输出
  bottlenect_values = np.squeeze(bottlenect_values)
  return bottlenect_values