Tensor conversion requested dtype float32_ref for Tensor with dtype float32:

主要原因应该是TensorFlow版本造成的。修改placeholder后可以跑通了。

原代码：

self.tensor_input = tf.placeholder(tf.float32, shape=(None, 112, 112, 3), name='patch')

 

修改后：

 graph = tf.Graph()

with graph.as_default():
    self.tensor_input = tf.placeholder(tf.float32, shape=(None, 112, 112, 3), name='patch')