Python tensorflow model into PB file

最新推荐文章于 2022-03-29 17:01:25 发布
最新推荐文章于 2022-03-29 17:01:25 发布
阅读量174 收藏
点赞数
分类专栏： Python English blog
你若不肯低头，谁能为你戴上桂冠
本文链接：https://blog.csdn.net/m0_37690102/article/details/106981210
版权

使用python打印model模型及pb模型的输入输出节点。参考下面的代码：

  • The Input Output nodes of the Model model and THE PB model are printed in Python.Refer to the following code:
from tensorflow.python import pywrap_tensorflow
import os

checkpoint_path=os.path.join('checkpoints/crnn/models-19')
reader=pywrap_tensorflow.NewCheckpointReader(checkpoint_path)
var_to_shape_map=reader.get_variable_to_shape_map()
for key in var_to_shape_map:
    print('tensor_name: ',key)

output：

tensor_name:  vgg/conv3/conv3_2/BatchNorm/beta/Adam_1
tensor_name:  beta1_power
tensor_name:  vgg/conv4/conv4_2/BatchNorm/moving_variance
tensor_name:  fully_connected/fully_connected/weights/Adam_1
tensor_name:  vgg/conv1/conv1_1/BatchNorm/beta/Adam
tensor_name:  vgg/conv4/conv4_1/BatchNorm/moving_variance
tensor_name:  global_step
tensor_name:  vgg/conv3/conv3_1/BatchNorm/beta/Adam_1
tensor_name:  vgg/conv3/conv3_1/BatchNorm/beta/Adam
tensor_name:  vgg/conv1/conv1_1/weights/Adam_1
tensor_name:  vgg/conv1/conv1_1/weights/Adam
tensor_name:  beta2_power
tensor_name:  vgg/conv3/conv3_2/BatchNorm/beta/Adam
tensor_name:  fully_connected/fully_connected/weights/Adam
tensor_name:  stack_bidirectional_rnn/cell_0/bidirectional_rnn/bw/basic_lstm_cell/kernel
tensor_name:  fully_connected/fully_connected/biases
tensor_name:  fully_connected/fully_connected/biases/Adam_1
tensor_name:  fully_connected/fully_connected/biases/Adam
tensor_name:  vgg/conv3/conv3_2/BatchNorm/beta
tensor_name:  fully_connected/fully_connected/weights
tensor_name:  stack_bidirectional_rnn/cell_0/bidirectional_rnn/bw/basic_lstm_cell/bias
tensor_name:  stack_bidirectional_rnn/cell_0/bidirectional_rnn/bw/basic_lstm_cell/bias/Adam_1
tensor_name:  stack_bidirectional_rnn/cell_0/bidirectional_rnn/bw/basic_lstm_cell/bias/Adam
tensor_name:  vgg/conv3/conv3_2/weights/Adam_1
tensor_name:  stack_bidirectional_rnn/cell_0/bidirectional_rnn/bw/basic_lstm_cell/kernel/Adam
tensor_name:  vgg/conv3/conv3_2/weights/Adam
tensor_name:  stack_bidirectional_rnn/cell_0/bidirectional_rnn/bw/basic_lstm_cell/kernel/Adam_1
tensor_name:  stack_bidirectional_rnn/cell_0/bidirectional_rnn/fw/basic_lstm_cell/bias
tensor_name:  vgg/conv3/conv3_2/BatchNorm/moving_mean
tensor_name:  stack_bidirectional_rnn/cell_0/bidirectional_rnn/fw/basic_lstm_cell/bias/Adam
tensor_name:  stack_bidirectional_rnn/cell_0/bidirectional_rnn/fw/basic_lstm_cell/bias/Adam_1
tensor_name:  stack_bidirectional_rnn/cell_0/bidirectional_rnn/fw/basic_lstm_cell/kernel
tensor_name:  stack_bidirectional_rnn/cell_0/bidirectional_rnn/fw/basic_lstm_cell/kernel/Adam
tensor_name:  stack_bidirectional_rnn/cell_0/bidirectional_rnn/fw/basic_lstm_cell/kernel/Adam_1
tensor_name:  stack_bidirectional_rnn/cell_1/bidirectional_rnn/fw/basic_lstm_cell/kernel/Adam_1
tensor_name:  stack_bidirectional_rnn/cell_1/bidirectional_rnn/bw/basic_lstm_cell/bias
tensor_name:  vgg/conv2/conv2_1/BatchNorm/moving_mean
tensor_name:  stack_bidirectional_rnn/cell_1/bidirectional_rnn/bw/basic_lstm_cell/kernel
tensor_name:  stack_bidirectional_rnn/cell_1/bidirectional_rnn/bw/basic_lstm_cell/bias/Adam
tensor_name:  stack_bidirectional_rnn/cell_1/bidirectional_rnn/fw/basic_lstm_cell/kernel
tensor_name:  stack_bidirectional_rnn/cell_1/bidirectional_rnn/bw/basic_lstm_cell/bias/Adam_1
tensor_name:  vgg/conv1/conv1_1/BatchNorm/moving_mean
tensor_name:  stack_bidirectional_rnn/cell_1/bidirectional_rnn/bw/basic_lstm_cell/kernel/Adam
tensor_name:  stack_bidirectional_rnn/cell_1/bidirectional_rnn/bw/basic_lstm_cell/kernel/Adam_1
tensor_name:  stack_bidirectional_rnn/cell_1/bidirectional_rnn/fw/basic_lstm_cell/bias
tensor_name:  vgg/conv3/conv3_1/BatchNorm/moving_mean
tensor_name:  stack_bidirectional_rnn/cell_1/bidirectional_rnn/fw/basic_lstm_cell/bias/Adam
tensor_name:  stack_bidirectional_rnn/cell_1/bidirectional_rnn/fw/basic_lstm_cell/kernel/Adam
tensor_name:  stack_bidirectional_rnn/cell_1/bidirectional_rnn/fw/basic_lstm_cell/bias/Adam_1
tensor_name:  vgg/conv1/conv1_1/BatchNorm/beta
tensor_name:  vgg/conv3/conv3_1/weights/Adam
tensor_name:  vgg/conv1/conv1_1/BatchNorm/beta/Adam_1
tensor_name:  vgg/conv2/conv2_1/BatchNorm/moving_variance
tensor_name:  vgg/conv3/conv3_1/BatchNorm/beta
tensor_name:  vgg/conv2/conv2_1/weights/Adam
tensor_name:  vgg/conv1/conv1_1/BatchNorm/moving_variance
tensor_name:  vgg/conv1/conv1_1/weights
tensor_name:  vgg/conv2/conv2_1/BatchNorm/beta
tensor_name:  vgg/conv2/conv2_1/BatchNorm/beta/Adam
tensor_name:  vgg/conv2/conv2_1/BatchNorm/beta/Adam_1
tensor_name:  vgg/conv3/conv3_1/weights/Adam_1
tensor_name:  vgg/conv2/conv2_1/weights
tensor_name:  vgg/conv4/conv4_1/BatchNorm/beta/Adam
tensor_name:  vgg/conv2/conv2_1/weights/Adam_1
tensor_name:  vgg/conv3/conv3_1/weights
tensor_name:  vgg/conv3/conv3_1/BatchNorm/moving_variance
tensor_name:  vgg/conv3/conv3_2/BatchNorm/moving_variance
tensor_name:  vgg/conv3/conv3_2/weights
tensor_name:  vgg/conv4/conv4_1/BatchNorm/beta
tensor_name:  vgg/conv4/conv4_1/BatchNorm/beta/Adam_1
tensor_name:  vgg/conv4/conv4_1/BatchNorm/moving_mean
tensor_name:  vgg/conv4/conv4_1/weights
tensor_name:  vgg/conv4/conv4_1/weights/Adam
tensor_name:  vgg/conv4/conv4_1/weights/Adam_1
tensor_name:  vgg/conv4/conv4_2/BatchNorm/beta
tensor_name:  vgg/conv4/conv4_2/BatchNorm/beta/Adam
tensor_name:  vgg/conv4/conv4_2/BatchNorm/beta/Adam_1
tensor_name:  vgg/conv4/conv4_2/BatchNorm/moving_mean
tensor_name:  vgg/conv4/conv4_2/weights
tensor_name:  vgg/conv4/conv4_2/weights/Adam
tensor_name:  vgg/conv4/conv4_2/weights/Adam_1
tensor_name:  vgg/conv5/conv5_1/BatchNorm/beta
tensor_name:  vgg/conv5/conv5_1/BatchNorm/beta/Adam
tensor_name:  vgg/conv5/conv5_1/BatchNorm/beta/Adam_1
tensor_name:  vgg/conv5/conv5_1/BatchNorm/moving_mean
tensor_name:  vgg/conv5/conv5_1/BatchNorm/moving_variance
tensor_name:  vgg/conv5/conv5_1/weights
tensor_name:  vgg/conv5/conv5_1/weights/Adam
tensor_name:  vgg/conv5/conv5_1/weights/Adam_1

打印pb文件网络输入输出节点：

  • Network Input/output node of PB file printing:

import tensorflow as tf
import os

model_name = './tmp/frozen_model_crnn.pb'

def create_graph():
    with tf.gfile.FastGFile(model_name, 'rb') as f:
        graph_def = tf.GraphDef()
        graph_def.ParseFromString(f.read())
        tf.import_graph_def(graph_def, name='')

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')

output：

Placeholder 

vgg/conv1/conv1_1/weights 

vgg/conv1/conv1_1/weights/read 

vgg/conv1/conv1_1/Conv2D 

vgg/conv1/conv1_1/BatchNorm/Const 

vgg/conv1/conv1_1/BatchNorm/beta 

vgg/conv1/conv1_1/BatchNorm/beta/read 

vgg/conv1/conv1_1/BatchNorm/Const_1 

vgg/conv1/conv1_1/BatchNorm/Const_2 

vgg/conv1/conv1_1/BatchNorm/FusedBatchNorm 

vgg/conv1/conv1_1/Relu 

vgg/pool1/MaxPool 

vgg/conv2/conv2_1/weights 

vgg/conv2/conv2_1/weights/read 

vgg/conv2/conv2_1/Conv2D 

vgg/conv2/conv2_1/BatchNorm/Const 

vgg/conv2/conv2_1/BatchNorm/beta 

vgg/conv2/conv2_1/BatchNorm/beta/read 

vgg/conv2/conv2_1/BatchNorm/Const_1 

vgg/conv2/conv2_1/BatchNorm/Const_2 

vgg/conv2/conv2_1/BatchNorm/FusedBatchNorm 

vgg/conv2/conv2_1/Relu 

vgg/pool2/MaxPool 

vgg/conv3/conv3_1/weights 

vgg/conv3/conv3_1/weights/read 

vgg/conv3/conv3_1/Conv2D 

vgg/conv3/conv3_1/BatchNorm/Const 

vgg/conv3/conv3_1/BatchNorm/beta 

vgg/conv3/conv3_1/BatchNorm/beta/read 

vgg/conv3/conv3_1/BatchNorm/Const_1 

vgg/conv3/conv3_1/BatchNorm/Const_2 

vgg/conv3/conv3_1/BatchNorm/FusedBatchNorm 

vgg/conv3/conv3_1/Relu 

vgg/conv3/conv3_2/weights 

vgg/conv3/conv3_2/weights/read 

vgg/conv3/conv3_2/Conv2D 

vgg/conv3/conv3_2/BatchNorm/Const 

vgg/conv3/conv3_2/BatchNorm/beta 

vgg/conv3/conv3_2/BatchNorm/beta/read 

vgg/conv3/conv3_2/BatchNorm/Const_1 

vgg/conv3/conv3_2/BatchNorm/Const_2 

vgg/conv3/conv3_2/BatchNorm/FusedBatchNorm 

vgg/conv3/conv3_2/Relu 

vgg/pool3/MaxPool 

vgg/conv4/conv4_1/weights 

vgg/conv4/conv4_1/weights/read 

vgg/conv4/conv4_1/Conv2D 

vgg/conv4/conv4_1/BatchNorm/Const 

vgg/conv4/conv4_1/BatchNorm/beta 

vgg/conv4/conv4_1/BatchNorm/beta/read 

vgg/conv4/conv4_1/BatchNorm/Const_1 

vgg/conv4/conv4_1/BatchNorm/Const_2 

vgg/conv4/conv4_1/BatchNorm/FusedBatchNorm 

vgg/conv4/conv4_1/Relu 

vgg/conv4/conv4_2/weights 

vgg/conv4/conv4_2/weights/read 

vgg/conv4/conv4_2/Conv2D 

vgg/conv4/conv4_2/BatchNorm/Const 

vgg/conv4/conv4_2/BatchNorm/beta 

vgg/conv4/conv4_2/BatchNorm/beta/read 

vgg/conv4/conv4_2/BatchNorm/Const_1 

vgg/conv4/conv4_2/BatchNorm/Const_2 

vgg/conv4/conv4_2/BatchNorm/FusedBatchNorm 

vgg/conv4/conv4_2/Relu 

vgg/pool4/MaxPool 

vgg/conv5/conv5_1/weights 

vgg/conv5/conv5_1/weights/read 

vgg/conv5/conv5_1/Conv2D 

vgg/conv5/conv5_1/BatchNorm/Const 

vgg/conv5/conv5_1/BatchNorm/beta 

vgg/conv5/conv5_1/BatchNorm/beta/read 

vgg/conv5/conv5_1/BatchNorm/Const_1 

vgg/conv5/conv5_1/BatchNorm/Const_2 

vgg/conv5/conv5_1/BatchNorm/FusedBatchNorm 

vgg/conv5/conv5_1/Relu

CPT model to pb

"""
此文件可以把ckpt模型转为pb模型
"""
'''
tensor_name:  vgg/conv3/conv3_2/BatchNorm/beta/Adam_1
tensor_name:  beta1_power
tensor_name:  vgg/conv4/conv4_2/BatchNorm/moving_variance
tensor_name:  fully_connected/fully_connected/weights/Adam_1
tensor_name:  vgg/conv1/conv1_1/BatchNorm/beta/Adam
tensor_name:  vgg/conv4/conv4_1/BatchNorm/moving_variance
tensor_name:  global_step
tensor_name:  vgg/conv3/conv3_1/BatchNorm/beta/Adam_1
tensor_name:  vgg/conv3/conv3_1/BatchNorm/beta/Adam
tensor_name:  vgg/conv1/conv1_1/weights/Adam_1
tensor_name:  vgg/conv1/conv1_1/weights/Adam
tensor_name:  beta2_power
tensor_name:  vgg/conv3/conv3_2/BatchNorm/beta/Adam
tensor_name:  fully_connected/fully_connected/weights/Adam
tensor_name:  stack_bidirectional_rnn/cell_0/bidirectional_rnn/bw/basic_lstm_cell/kernel
tensor_name:  fully_connected/fully_connected/biases
tensor_name:  fully_connected/fully_connected/biases/Adam_1
tensor_name:  fully_connected/fully_connected/biases/Adam
tensor_name:  vgg/conv3/conv3_2/BatchNorm/beta
tensor_name:  fully_connected/fully_connected/weights
tensor_name:  stack_bidirectional_rnn/cell_0/bidirectional_rnn/bw/basic_lstm_cell/bias
tensor_name:  stack_bidirectional_rnn/cell_0/bidirectional_rnn/bw/basic_lstm_cell/bias/Adam_1
tensor_name:  stack_bidirectional_rnn/cell_0/bidirectional_rnn/bw/basic_lstm_cell/bias/Adam
tensor_name:  vgg/conv3/conv3_2/weights/Adam_1
tensor_name:  stack_bidirectional_rnn/cell_0/bidirectional_rnn/bw/basic_lstm_cell/kernel/Adam
tensor_name:  vgg/conv3/conv3_2/weights/Adam
tensor_name:  stack_bidirectional_rnn/cell_0/bidirectional_rnn/bw/basic_lstm_cell/kernel/Adam_1
tensor_name:  stack_bidirectional_rnn/cell_0/bidirectional_rnn/fw/basic_lstm_cell/bias
tensor_name:  vgg/conv3/conv3_2/BatchNorm/moving_mean
tensor_name:  stack_bidirectional_rnn/cell_0/bidirectional_rnn/fw/basic_lstm_cell/bias/Adam
tensor_name:  stack_bidirectional_rnn/cell_0/bidirectional_rnn/fw/basic_lstm_cell/bias/Adam_1
tensor_name:  stack_bidirectional_rnn/cell_0/bidirectional_rnn/fw/basic_lstm_cell/kernel
tensor_name:  stack_bidirectional_rnn/cell_0/bi
'''
#唯一要改动的就是output_node_names，指定的最后一层输出节点名称，这个是你自己设定的
import tensorflow as tf
#from create_tf_record import *
from tensorflow.python.framework import graph_util

def freeze_graph(input_checkpoint,output_graph):
    '''
    :param input_checkpoint:
    :param output_graph: PB模型保存路径
    :return:
    '''
    # checkpoint = tf.train.get_checkpoint_state(model_folder) #检查目录下ckpt文件状态是否可用
    # input_checkpoint = checkpoint.model_checkpoint_path #得ckpt文件路径
 
    # 指定输出的节点名称,该节点名称必须是原模型中存在的节点
    # 直接用最后输出的节点，可以在tensorboard中查找到，tensorboard只能在linux中使用

    saver = tf.train.import_meta_graph(input_checkpoint + '.meta', clear_devices=True)
    graph = tf.get_default_graph() # 获得默认的图
    input_graph_def = graph.as_graph_def()  # 返回一个序列化的图代表当前的图
 
    with tf.Session() as sess:
        saver.restore(sess, input_checkpoint) #恢复图并得到数据
        output_graph_def = graph_util.convert_variables_to_constants(  # 模型持久化，将变量值固定
            sess=sess,
            input_graph_def=input_graph_def,# 等于:sess.graph_def
            output_node_names=output_node_names.split(","))# 如果有多个输出节点，以逗号隔开
 
        with tf.gfile.GFile(output_graph, "wb") as f: #保存模型
            f.write(output_graph_def.SerializeToString()) #序列化输出
        print("%d ops in the final graph." % len(output_graph_def.node)) #得到当前图有几个操作节点



#========================================================================================
input_checkpoint='checkpoints/crnn/models-19' #crnn
out_pb_path='checkpoints/frozen_model_crnn.pb'
output_node_names = "vgg/conv5/conv5_1/Relu"
#========================================================================================



freeze_graph(input_checkpoint, out_pb_path)

 I hope I can help you,If you have any questions, please  comment on this blog or send me a private message. I will reply in my free time.

李伯爵的指间沙
关注
  • 0
    点赞
  • 踩
  • 0
    收藏
    觉得还不错? 一键收藏
  • 0
    评论
专栏目录
(B20_01) 查看pb模型的输入输出节点
jn10010537的博客
03-02 4122
在进行pb模型文件量化，比如转tflite的时候，或是推理的时候，往往需要知道该模型文件的输入输出节点信息。 #coding:utf-8 #方法1，获得输入、输出的node结构 import tensorflow as tf #_________获取python的版本号________# import sys print('>>>Python Version:',sys.v...
ckpt文件生成固化的PB文件并查看模型的输入与输出
向高手学习
07-10 2598
概述 需要从ckpt文件生成固化的PB文件，给生成现场用。现在我探索下不依赖代码里的网络结构、仅仅用CKPT文件来生成PB文件。 转化 训练生成的ckpt文件目录如下： $ ls ./ckpt0507/ checkpoint graph.pbtxt model.ckpt-3251.data-...
如何用Tensorflow训练模型成pb文件（一）——基于原始图片的读取
Alex_XT的博客
09-09 9912
简介为何要生成pb文件，大家应该有所了解吧，这里是提供Android的调用，即将Tensorflow训练好了的模型结构和参数移植到Android手机上。训练读取原始图片过程，将其ratio=0.2为校验样本，0.8的比重为训练样本。设置图片宽w = 200，高h = 150，通道c=3，类别数量n_classes = 2。这里提供了transform.resize(img, (h, w, c))将不
查看tf serving使用的pb模型的输入输出
samoyan的博客,记录技术成长~
03-29 2868
使用以下命令： saved_model_cli show --all --dir output_pb/ 输出如下： saved_model_cli show --all --dir output_pb/1648525103/ 2022-03-29 16:59:55.132993: I tensorflow/stream_executor/platform/default/dso_loader.cc:49] Successfully opened dynamic library libcuda
TensorFlow查看pb模型节点信息
LuchangLi 的专栏
08-22 2351
example: import tensorflow as tf import os model_dir = './' model_name = 'ocr.pb' def create_graph(): with tf.gfile.FastGFile(os.path.join( model_dir, model_name), 'rb') as f: graph_def = tf.GraphDef() graph_def.ParseFromS.
python 解析pb文件_将tensorflow模型打包成PB文件及PB文件读取方式
weixin_42365141的博客
12-29 1634
1. tensorflow模型文件打包成PB文件import tensorflow as tffrom tensorflow.python.tools import freeze_graphwith tf.Graph().as_default():with tf.device("/cpu:0"):config = tf.ConfigProto(allow_soft_placement=True)w...
tensorflow保存模型为pb文件的各种方式
shuijinghua的博客
07-05 7822
*.pb,官方描述如下: GraphDef(.pb)-a protobuf that represents the Tensorflow training and or computation graph. This contains operators, tensors, and variables definitions. FrozenGraphDef - a subclass of Gr...
机器学习 tensorflow 2 的model 的保存，导入和继续培训
曾立文的博客
11-17 2085
关于保存和导入可以参看https://www.tensorflow.org/guide/keras/save_and_serialize # Save the model model.save('path_to_my_model.h5') # Recreate the exact same model purely from the file new_model = keras.model...
python中tensorflow实现二分类_利用Tensorflow，神经网络进行二类分类
weixin_36160268的博客
01-29 809
我正在尝试使用Tensorflow对图像进行两类分类。在我想随机抽取1000个像素。但是，我遇到了错误。在"logits = inference(images_placeholder, keep_prob)File "train5.py", line 83, in inferencelist = random.sample(x_image(IMAGE_PIXELS),SAMPLE_PIXELS)T...
人工智能python3+tensorflow人脸识别_机器学习tensorflow object detection 实现人脸识别...
weixin_36068140的博客
02-22 501
object detection是Tensorflow很常用的api，功能强大，很有想象空间，人脸识别，花草识别，物品识别等。下面是我做实验的全过程，使用自己收集的胡歌图片，实现人脸识别，找出胡歌。安装tensorflow官方的教程已经写得非常好了，这里就不多说，但是有一点必须注意的是，必须安装python3.6版本，不能安装最新的python3.7版本，不然会出现很多不兼容的问题难以处理。尽可能...
TensorFlow查看输入节点和输出节点名称方式
12-23
TensorFlow 定义输入节点名称input_name： with tf.name_scope('input'): bottleneck_input = tf.placeholder_with_default( bottleneck_tensor, shape=[batch_size, bottleneck_tensor_size], name='Mul') TensorFlow查看pb数据库里面的输入节点和输出节点： import tensorflow as tf import os model_dir = './tmp/' model_name =
[Tensorflow] 如何从pb文件生成标准的tensorflow checkpoint文件？
Hazekiah的博客
10-30 2116
How to generate a tensorflow checkpoint file given a pb file?
tensorflow的pb模型转为onnx模型
weixin_43836242的博客
01-28 1万+
pb模型转为onnx模型
tensorboard ckpt pb 模型的输出节点_FlyAI小课堂：Tensorflow-模型和数据的保存和载入...
weixin_39645343的博客
11-22 211
在开始学习之前推荐大家可以多在FlyAI竞赛服务平台多参加训练和竞赛，以此来提升自己的能力。FlyAI是为AI开发者提供数据竞赛并支持GPU离线训练的一站式服务平台。每周免费提供项目开源算法样例，支持算法能力变现以及快速的迭代算法模型。目录基本方法不需重新定义网络结构的方法saved_model方式附件一：sklearn上的用法一、基本方法1.1 保存定义变量使用saver.save()方法保存i...
Tensorflow Summary: 查看Tensorflow Model pb格式模型的信息
wishchinYang的专栏
05-08 5752
参考文链:如何查看Tensorflow SavedModel格式模型的信息; 参考:summary方法--小酒窝的博客; 参考:tensorflow中输出参数的方法--详细; tensorflow生成的pb图,netron嫌弃太大打不开, 因此需要借助于tensorboard查看工具; 0.查看模型的Signature签名 这里的签名，并非是为了保证模型不被修改的那种电子签名。我的...
查看tensorflow 模型文件的节点信息
热门推荐
haima1998的专栏
05-13 1万+
转自：https://www.cnblogs.com/bonelee/p/8462578.html查看tensorflow pb模型文件的节点信息：import tensorflow as tf with tf.Session() as sess: with open('./quantized_model.pb', 'rb') as f: graph_def = tf.Gr...
CNN: 查看tensorflow Pb模型所有层的名字
wishchinYang的专栏
05-28 4876
tensorflow的keras没有外部可视化配置文件,需要使用netron查看模型参数,或者从程序中打印出来, 打印代码: def get_all_layernames(): """get all layers name""" model_dir = os.path.join(ROOT_DIR, 'stru/') pb_file_path = os.path.jo...
打印pb文件的输入输出结点信息
qq798446835的博客
01-13 1287
import os import tensorflow as tf model_dir = r'E:\alrogirths\models\research\object_detection\myself\saved_model' model_name = 'frozen_inference_graph.pb' def create_graph(): with tf.gfile.FastGF...
tensorflow框架.ckpt .pb模型节点tensor_name打印及ckpt模型转.pb模型
嘿芝麻的树洞
08-29 1万+
转换模型首先要知道的是从哪个节点输出，如果没有源代码是很难清楚节点信息。 获取ckpt模型的节点名称 import os from tensorflow.python import pywrap_tensorflow checkpoint_path = os.path.join('./ade20k', "model.ckpt-27150") reader = pywrap_tensorf...
Python TensorFlow实现MNIST手写数字识别教程
"该资源是一个关于手写体数字识别的项目，使用了Python编程语言以及TensorFlow深度学习框架。项目基于MNIST数据集，该数据集含有60000个训练样本和10000个测试样本，用于训练全连接的神经网络。MNIST_model文件夹...
评论
添加红包

请填写红包祝福语或标题

个

红包个数最小为10个

元

红包金额最低5元

当前余额3.43元 前往充值 >
需支付：10.00元
成就一亿技术人!
领取后你会自动成为博主和红包主的粉丝 规则
hope_wisdom
发出的红包
实付元
使用余额支付
点击重新获取
扫码支付
钱包余额 0

抵扣说明：

1.余额是钱包充值的虚拟货币，按照1:1的比例进行支付金额的抵扣。
2.余额无法直接购买下载，可以购买VIP、付费专栏及课程。

余额充值