将TensorFlow CKPT模型权重保存为Numpy数组的方法

小石头

发布时间：2026-10-13| 6人浏览过

原创

本文介绍如何使用Python将TensorFlow的CKPT模型文件转换为Numpy数组格式。通过pywrap_tensorflow读取检查点，解析变量名并提取权重和偏置，最终保存为.npy文件，便于后续分析或部署使用。

具体实现代码示例如下：

#coding=gbk
import numpy as np
import tensorflow as tf
from tensorflow.python import pywrap_tensorflow

checkpoint_path='model.ckpt-5000'#your ckpt path
reader=pywrap_tensorflow.NewCheckpointReader(checkpoint_path)
var_to_shape_map=reader.get_variable_to_shape_map()

alexnet={}
alexnet_layer = ['conv1','conv2','conv3','conv4','conv5','fc6','fc7','fc8']
add_info = ['weights','biases']

alexnet={'conv1':[[],[]],'conv2':[[],[]],'conv3':[[],[]],'conv4':[[],[]],'conv5':[[],[]],'fc6':[[],[]],'fc7':[[],[]],'fc8':[[],[]]}


for key in var_to_shape_map:
 #print ("tensor_name",key)

 str_name = key
 # 因为模型使用Adam算法优化的，在生成的ckpt中，有Adam后缀的tensor
 if str_name.find('Adam') > -1:
  continue

 print('tensor_name:' , str_name)

 if str_name.find('/') > -1:
  names = str_name.split('/')
  # first layer name and weight, bias
  layer_name = names[0]
  layer_add_info = names[1]
 else:
  layer_name = str_name
  layer_add_info = None

 if layer_add_info == 'weights':
  alexnet[layer_name][0]=reader.get_tensor(key)
 elif layer_add_info == 'biases':
  alexnet[layer_name][1] = reader.get_tensor(key)
 else:
  alexnet[layer_name] = reader.get_tensor(key)

# save npy
np.save('alexnet_pointing04.npy',alexnet)
print('save npy over...')
#print(alexnet['conv1'][0].shape)
#print(alexnet['conv1'][1].shape)

以上就是将TensorFlow CKPT模型存储为Numpy格式的全部内容，希望能为读者提供参考，同时也欢迎大家继续支持本站。