TensorFlow獲取加載模型中的全部張量名稱

 核心代碼如下：

[tensor.name for tensor in tf.get_default_graph().as_graph_def().node]

---

實例代碼：（加載了Inceptino_v3的模型，並獲取該模型所有節點的名稱）

# -*- coding: utf-8 -*-

import tensorflow as tf
import os

model_dir = 'C:/Inception_v3'
model_name = 'output_graph.pb'

# 讀取並創建一個圖graph來存放訓練好的 Inception_v3模型（函數）
def create_graph():
    with tf.gfile.FastGFile(os.path.join(
            model_dir, model_name), 'rb') as f:
        # 使用tf.GraphDef()定義一個空的Graph
        graph_def = tf.GraphDef()
        graph_def.ParseFromString(f.read())
        # Imports the graph from graph_def into the current default Graph.
        tf.import_graph_def(graph_def, name='')

# 創建graph
create_graph()

tensor_name_list = [tensor.name for tensor in tf.get_default_graph().as_graph_def().node]
for tensor_name in tensor_name_list:
    print(tensor_name,'\n')

---

輸出結果：

mixed_8/tower/conv_1/batchnorm/moving_variance 

mixed_8/tower/conv_1/batchnorm 

r_1/mixed/conv_1/batchnorm 

.

.

.

mixed_10/tower_1/mixed/conv_1/CheckNumerics 

mixed_10/tower_1/mixed/conv_1/control_dependency 

mixed_10/tower_1/mixed/conv_1 

pool_3 

pool_3/_reshape/shape 

pool_3/_reshape 

input/BottleneckInputPlaceholder 
.
.
.
.
final_training_ops/weights/final_weights 

final_training_ops/weights/final_weights/read 

final_training_ops/biases/final_biases 

final_training_ops/biases/final_biases/read 

final_training_ops/Wx_plus_b/MatMul 

final_training_ops/Wx_plus_b/add 

final_result

由於結果太長了，就省略了一些。

---

如果不想這樣print輸出也可以將其寫入txt 查看。

寫入txt代碼如下：

tensor_name_list = [tensor.name for tensor in tf.get_default_graph().as_graph_def().node]

txt_path = './txt/節點名稱'
full_path = txt_path+ '.txt'

for tensor_name in tensor_name_list:
    name = tensor_name + '\n'
    file = open(full_path,'a+')
file.write(name)
file.close()

參考鏈接：

TensorFlow學習筆記：獲取以來模型全部張量名稱

Tensorflow：如何通過名稱獲得張量？