tensorflow pb模型获得权重wts

最新推荐文章于 2024-07-25 11:18:26 发布

原创最新推荐文章于 2024-07-25 11:18:26 发布 · 905 阅读

1 ·

CC 4.0 BY-SA版权

文章标签：

#python #tensorflow #深度学习

该博客主要展示了如何使用TensorFlow加载.pb模型文件，并提取其中的权重。通过读取模型定义，打印出节点名称及其对应的权重值，揭示了模型内部的参数细节。

部署运行你感兴趣的模型镜像

tensorflow pb模型获得权重wts

import tensorflow as tf
from tensorflow.python.platform import gfile
#path to your .pb file
GRAPH_PB_PATH = './model/tensorflow_inception_v3_stripped_optimized_quantized.pb' 
with tf.Session(config=config) as sess:
  print("load graph")
  with gfile.FastGFile(GRAPH_PB_PATH,'rb') as f:
    graph_def = tf.GraphDef()
    graph_def.ParseFromString(f.read())
    sess.graph.as_default()
    tf.import_graph_def(graph_def, name='')
    graph_nodes=[n for n in graph_def.node]
wts = [n for n in graph_nodes if n.op=='Const']
from tensorflow.python.framework import tensor_util
 
for n in wts:
    print("Name of the node - %s" % n.name)
    print("Value - ")
    print(tensor_util.MakeNdarray(n.attr['value'].tensor))

运行出现报错

Traceback (most recent call last):
  File "gen_pb_wts.py", line 5, in <module>
    with tf.Session(config=config) as sess:
NameError: name 'config' is not defined

将with tf.Session(config=config) as sess: 修改为tf.Session(）as sess: 重新运行部分结果如下

Name of the node - BoxPredictor_4/ClassPredictor_depthwise/BatchNorm/gamma
Value -
[1.0076598  0.600361   0.64767396 0.5886829  0.6474589  0.57899725
 0.7420581  0.43999094 0.9291541  0.6562006  0.6434446  0.62888956
 0.7723971  1.013845   0.5593387  0.5763445  0.60916996 0.66238034
 1.0512784  0.58268934 0.52825654 0.55082077 0.4491668  0.6486317
 0.717447   0.9524743  0.7800959  0.42979443 0.55822694 0.60574436
 1.0343778  0.5154323  0.85758215 1.0833062  0.5687403  0.5752864
 0.84560686 0.62819284 0.6473139  0.58606607 0.6454761  0.6270443
 0.5428621  0.673155   0.9136668  0.62957346 0.855056   0.80063677
 0.55059046 0.941908   0.6897346  0.5215224  0.64170074 1.0129
 0.63292354 0.7850904  0.44446874 0.7453388  0.6934488  0.5419593
 0.70664483 0.48354548 0.60087705 0.6893038  0.45109886 0.62180865
 0.6481066  0.67009485 0.6451158  0.5812356  0.8095073  0.40086615
 0.50798595 0.49360913 0.51697165 0.5107985  0.53685313 0.9088723
 0.76556355 0.36509857 0.367629   0.80191237 0.6855312  0.970307
 1.0197186  0.671655   0.57405996 0.76866794 0.74245185 0.51531
 0.6980881  0.8520622  0.54880375 0.91214556 0.710224   0.41681576
 0.6096677  0.36241293 0.56075    0.5863032  0.6354018  0.5120003
 1.0113248  0.8348537  0.7205713  0.67849016 0.97358465 0.44886646
 0.8159403  0.64450437 0.5997353  0.54739535 0.6758225  0.61986154
 0.6395207  0.5936247  0.5463965  0.5821391  0.6760462  1.0680252
 0.55302185 0.7000822  0.8675259  0.6098902  0.74904156 0.5757958
 0.2752119  0.64663106 0.56606585 0.5852428  0.82287776 0.5280436
 0.6185344  0.6950132  1.0274562  0.7259649  1.0736766  0.602673
 0.8857293  0.59187907 0.78937036 0.4410297  0.7101182  0.8159739
 0.55530983 0.6026916  0.7655125  0.63132054 0.52071196 0.57674795
 0.919869   0.5550313  0.55495024 1.0107807  0.6631691  0.7698098
 0.6434384  0.6741949  1.0766468  0.8970208  0.6370691  0.48358867
 0.9154165  0.5827703  0.9874309  1.0530525  0.600562   0.8497894
 0.6070317  0.5207939  0.7410717  0.75982136 0.57712764 0.539803
 0.72110265 1.1149868  0.6521509  0.86586046 0.49901482 0.6168051
 0.52331764 0.65172225 1.0159032  0.6780148  0.51903594 0.84838635
 0.6718971  0.8696038  0.4443711  0.48789653 0.7989498  0.6552638
 0.4824569  0.4769425  0.63836104 0.54260814 0.49502477 0.808891
 0.7465959  0.586385   0.63334984 0.63899726 0.6576707  0.99794906
 0.8469274  0.72964466 0.76935375 0.7605871  0.7410051  0.6782605
 0.4853573  0.90893656 0.835856   0.6664103  0.6316711  0.71951354
 0.7788348  0.58414733 0.85619235 0.87070334 0.6793145  0.9028463
 0.7173049  0.64837295 0.44326815 0.82141757 0.65800077 0.8714189
 0.90920067 0.7825849  0.5547908  1.0501481  0.6167052  0.80019474
 0.56159836 0.7303601  0.6019257  0.9394907  0.57830775 0.6206071