基于Python神经网络的手写数字识别实战与代码解析

Ada

发布时间：2026-10-13 | 4人浏览过

原创

本文基于《Python神经网络编程》一书，详细解析了使用Python构建简单神经网络以识别手写数字的实现过程。内容涵盖从图像像素读取、数据预处理（如灰度值取反与归一化）到模型训练及测试统计的完整代码逻辑。同时，文章探讨了针对特定像素图片训练的局限性，并分析了识别模糊现象的可能原因，为初学者提供了一份可直接参考的实践指南。

Python神经网络手写数字识别概述

本文分享了一个使用Python实现手写数字识别的具体代码示例，供读者参考学习。

神经网络核心类定义


import numpy
import scipy.special
#import matplotlib.pyplot
 
class neuralNetwork:
  def __init__(self,inputnodes,hiddennodes,outputnodes,learningrate):
    self.inodes=inputnodes
    self.hnodes=hiddennodes
    self.onodes=outputnodes
    
    self.lr=learningrate
  
    self.wih=numpy.random.normal(0.0,pow(self.hnodes,-0.5),(self.hnodes,self.inodes))
    self.who=numpy.random.normal(0.0,pow(self.onodes,-0.5),(self.onodes,self.hnodes))
    
    self.activation_function=lambda x: scipy.special.expit(x)
    pass
  
  def train(self,inputs_list,targets_list):
    inputs=numpy.array(inputs_list,ndmin=2).T
    targets=numpy.array(targets_list,ndmin=2).T
    
    hidden_inputs=numpy.dot(self.wih,inputs)
    hidden_outputs=self.activation_function(hidden_inputs)
    
    final_inputs=numpy.dot(self.who,hidden_outputs)
    final_outputs=self.activation_function(final_inputs)
    
    output_errors=targets-final_outputs
    hidden_errors=numpy.dot(self.who.T,output_errors)
    
    self.who+=self.lr*numpy.dot((output_errors*final_outputs*(1.0-final_outputs)),numpy.transpose(hidden_outputs))
    self.wih+=self.lr*numpy.dot((hidden_errors*hidden_outputs*(1.0-hidden_outputs)),numpy.transpose(inputs))
    pass
  
  def query(self,input_list):
    inputs=numpy.array(input_list,ndmin=2).T
    
    hidden_inputs=numpy.dot(self.wih,inputs)
    hidden_outputs=self.activation_function(hidden_inputs)
    
    final_inputs=numpy.dot(self.who,hidden_outputs)
    final_outputs=self.activation_function(final_inputs)
    
    return final_outputs
 
 
input_nodes=784
hidden_nodes=100
output_nodes=10
learning_rate=0.1
n=neuralNetwork(input_nodes,hidden_nodes,output_nodes,learning_rate)
 
training_data_file=open(r"C:\Users\lsy\Desktop\nn\mnist_train.csv","r")
training_data_list=training_data_file.readlines()
training_data_file.close()
#print(n.wih)
#print("")
epochs=2
for e in range(epochs):
  for record in training_data_list:
    all_values=record.split(",")
    inputs=(numpy.asfarray(all_values[1:])/255.0*0.99)+0.01
    targets=numpy.zeros(output_nodes)+0.01
    targets[int(all_values[0])]=0.99
    n.train(inputs,targets)
  
#print(n.wih)
#print(len(training_data_list))
#for i in training_data_list:
#  print(i)
 
test_data_file=open(r"C:\Users\lsy\Desktop\nn\mnist_test.csv","r")
test_data_list=test_data_file.readlines()
test_data_file.close()
 
scorecard=[]
 
 
for record in test_data_list:
  all_values=record.split(",")
  correct_lable=int(all_values[0])
  inputs=(numpy.asfarray(all_values[1:])/255.0*0.99)+0.01
  outputs=n.query(inputs)
  label=numpy.argmax(outputs)
  if(label==correct_lable):
    scorecard.append(1)
  else:
    scorecard.append(0)
 
scorecard_array=numpy.asarray(scorecard)
print(scorecard_array)
print("")
print(scorecard_array.sum()/scorecard_array.size)
#all_value=test_data_list[0].split(",")
#input=(numpy.asfarray(all_value[1:])/255.0*0.99)+0.01
#print(all_value[0])
 
#image_array=numpy.asfarray(all_value[1:]).reshape((28,28))
 
#matplotlib.pyplot.imshow(image_array,cmap="Greys",interpolation="None")
#matplotlib.pyplot.show()
#nn=n.query((numpy.asfarray(all_value[1:])/255.0*0.99)+0.01)
#for i in nn :
#  print(i)

以下代码源自《Python神经网络编程》一书，主要作为个人学习记录，方便日后查阅。

图像数据预处理与读取


image_file_name=r"*.JPG"
img_array=scipy.misc.imread(image_file_name,flatten=True)
 
img_data=255.0-img_array.reshape(784)
image_data=(img_data/255.0*0.99)+0.01

在图像数据处理阶段，需要读取图片对应的像素信息。由于训练集中灰度值的分布与实际显示相反，因此代码中采用255减去原始值的方式进行取反处理。

完整训练与测试代码实现


import numpy
import scipy.special
#import matplotlib.pyplot
import scipy.misc
from PIL import Image
class neuralNetwork:
  def __init__(self,inputnodes,hiddennodes,outputnodes,learningrate):
    self.inodes=inputnodes
    self.hnodes=hiddennodes
    self.onodes=outputnodes
    
    self.lr=learningrate
  
    self.wih=numpy.random.normal(0.0,pow(self.hnodes,-0.5),(self.hnodes,self.inodes))
    self.who=numpy.random.normal(0.0,pow(self.onodes,-0.5),(self.onodes,self.hnodes))
    
    self.activation_function=lambda x: scipy.special.expit(x)
    pass
  
  def train(self,inputs_list,targets_list):
    inputs=numpy.array(inputs_list,ndmin=2).T
    targets=numpy.array(targets_list,ndmin=2).T
    
    hidden_inputs=numpy.dot(self.wih,inputs)
    hidden_outputs=self.activation_function(hidden_inputs)
    
    final_inputs=numpy.dot(self.who,hidden_outputs)
    final_outputs=self.activation_function(final_inputs)
    
    output_errors=targets-final_outputs
    hidden_errors=numpy.dot(self.who.T,output_errors)
    
    self.who+=self.lr*numpy.dot((output_errors*final_outputs*(1.0-final_outputs)),numpy.transpose(hidden_outputs))
    self.wih+=self.lr*numpy.dot((hidden_errors*hidden_outputs*(1.0-hidden_outputs)),numpy.transpose(inputs))
    pass
  
  def query(self,input_list):
    inputs=numpy.array(input_list,ndmin=2).T
    
    hidden_inputs=numpy.dot(self.wih,inputs)
    hidden_outputs=self.activation_function(hidden_inputs)
    
    final_inputs=numpy.dot(self.who,hidden_outputs)
    final_outputs=self.activation_function(final_inputs)
    
    return final_outputs
 
 
input_nodes=784
hidden_nodes=100
output_nodes=10
learning_rate=0.1
n=neuralNetwork(input_nodes,hidden_nodes,output_nodes,learning_rate)
 
training_data_file=open(r"C:\Users\lsy\Desktop\nn\mnist_train.csv","r")
training_data_list=training_data_file.readlines()
training_data_file.close()
#print(n.wih)
#print("")
 
#epochs=2
#for e in range(epochs):
for record in training_data_list:
  all_values=record.split(",")
  inputs=(numpy.asfarray(all_values[1:])/255.0*0.99)+0.01
  targets=numpy.zeros(output_nodes)+0.01
  targets[int(all_values[0])]=0.99
  n.train(inputs,targets)
 
#image_file_name=r"C:\Users\lsy\Desktop\nn\1000-1.JPG"
'''
img_array=scipy.misc.imread(image_file_name,flatten=True)
img_data=255.0-img_array.reshape(784)
image_data=(img_data/255.0*0.99)+0.01
#inputs=(numpy.asfarray(image_data)/255.0*0.99)+0.01
outputs=n.query(image_data)
label=numpy.argmax(outputs)
print(label)
'''
#print(n.wih)
#print(len(training_data_list))
#for i in training_data_list:
#  print(i)
 
test_data_file=open(r"C:\Users\lsy\Desktop\nn\mnist_test.csv","r")
 
test_data_list=test_data_file.readlines()
test_data_file.close()
 
scorecard=[]
 
total=[0,0,0,0,0,0,0,0,0,0]
rightsum=[0,0,0,0,0,0,0,0,0,0]
 
for record in test_data_list:
  all_values=record.split(",")
  correct_lable=int(all_values[0])
  inputs=(numpy.asfarray(all_values[1:])/255.0*0.99)+0.01
  outputs=n.query(inputs)
  label=numpy.argmax(outputs)
  total[correct_lable]+=1
  if(label==correct_lable):
    scorecard.append(1)
    rightsum[correct_lable]+=1
  else:
    scorecard.append(0)
 
scorecard_array=numpy.asarray(scorecard)
print(scorecard_array)
print("")
print(scorecard_array.sum()/scorecard_array.size)
print("")
print(total)
print(rightsum)
for i in range(10):
  print((rightsum[i]*1.0)/total[i])
 
#all_value=test_data_list[0].split(",")
#input=(numpy.asfarray(all_value[1:])/255.0*0.99)+0.01
#print(all_value[0])
 
#image_array=numpy.asfarray(all_value[1:]).reshape((28,28))
 
#matplotlib.pyplot.imshow(image_array,cmap="Greys",interpolation="None")
#matplotlib.pyplot.show()
#nn=n.query((numpy.asfarray(all_value[1:])/255.0*0.99)+0.01)
#for i in nn :
#  print(i)