Tensorflow Pb 文件的一些操作

载入文件

参考链接：https://medium.com/@alexkn15/tensorflow-save-model-for-use-in-java-or-c-ab351a708ee4

model_filename = './saved_model.pb'
with tf.gfile.GFile(model_filename, 'rb') as f:
    graph_def = tf.GraphDef()
    graph_def.ParseFromString(f.read())
    # tensorflow adds "import/" prefix to all tensors when imports graph definition, ex: "import/input:0"
    # so we explicitly tell tensorflow to use empty string -> name=""
    tf.import_graph_def(graph_def, name="")
print(tf.get_default_graph().get_operations())  # just print all operations for debug

除此以外，查看载入图中的节点信息

node_list = graph_def.node
##其数据结构类似
print node_list[-1]
##----------------------
name: "prediction"
op: "Reshape"
input: "score"
input: "prediction/shape"
attr {
    key: "T"
    value {
        type: DT_INT32
    }
}
attr {
    key: "Tshape"
    value {
        type: DT_INT32
    }
}

查看所有节点的名字就

for node in graph_def.node
    print node.name

如果要找出所有Placeholder就

placeholder_nodes = [node for node in graph_def.node if node.op == "Placeholder"]