这是初学神经网络做的一个玩具级实战项目，识别系统是非常简单的四层神经网络 本项目分为两个部分。 ANN.py文件定义了神经网络的框架，并定义了供程序调用的函数接口 main.py文件是主程序文件，使用tkinter创建了GUI界面进行人机交互 另外三个.pkl文件是训练得到的表现比较好的权重（正确率为0.9679），请将它们与程序放在同一目录下。程序将在初始化神经网络时读取。
ANN_for_training.py和main_for_training.py是另一个程序，用于训练网络得到权重数据
具体思路在程序里有比较详细的注释。 如果有任何改进意见，欢迎交流！
This is a toy-level practical project done by a beginner neural network, and the recognition system is a very simple four-layer neural network The program is divided into two parts： ANN.py file defines the framework of the neural network and defines the function interface for the program to call main.py file is the main program file that uses tkinter to create a GUI interface for human-computer interaction The other three .pkl files are the well-performing weights trained (0.9679 correctness), so put them in the same directory as the program. The program will read it when the neural network is initialized.
ANN_for_training.py and main_for_training.py are another program for training the network to get weight data
The specific ideas are commented in more detail in the program. If you havw any suggestions, welcome to massage to me!