在张量流模型上测试图像

Question

我正在尝试测试一个 tensorflow 模型，其中包含 10 张图像作为批量大小，并作为输出获得特定区域的分割。这是我用来加载模型的代码

def run(model_path, image_path,graph_path ,output_path):
with tf.Session() as sess:
  model_saver = tf.train.import_meta_graph(graph_path, clear_devices=True)
  model_saver.restore(sess,tf.train.latest_checkpoint(LOGDIR))
  sess.run(tf.global_variables_initializer())
  graph = tf.get_default_graph()
  x=tf.placeholder(tf.float32,shape=[10, 400, 600, 3])
  output = graph.get_tensor_by_name("prediction:0")
  print("Model restored.")
  print('Initialized')
  temp=[]
  import glob
  from scipy import misc
  for file in os.listdir(image_path):
    if file.endswith(".jpg"):
      img = misc.imread(image_path + file)
      img = img.astype('Float32')
      img = np.resize(img,(400,600,3))
      temp.append(img)
  prediction = sess.run(output, feed_dict={x:temp})
  cv2.imwrite(output_path, prediction * 255)

但我收到以下错误：

InvalidArgumentError (see above for traceback): You must feed a value for placeholder tensor 'x' with dtype float and shape [10,400,600,3]
     [[Node: x = Placeholder[dtype=DT_FLOAT, shape=[10,400,600,3], _device="/job:localhost/replica:0/task:0/gpu:0"]()]]
     [[Node: prediction/_265 = _Recv[client_terminated=false, recv_device="/job:localhost/replica:0/task:0/cpu:0", send_device="/job:localhost/replica:0/task:0/gpu:0", send_device_incarnation=1, tensor_name="edge_28_prediction", tensor_type=DT_INT32, _device="/job:localhost/replica:0/task:0/cpu:0"]()]]

有谁知道这个错误是什么，或者是否有另一种测试模型的方法。

Answer 1

我认为你在这里重新定义了 x。而是按名称提供它，因此删除 x=placeholder() 行并使用 {'x:0': temp} 作为提要字典。