Skip to content

Tommy-Bie/Handwriting-digits-classification-based-on-CNN

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

10 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Handwriting-digits-classification-based-on-CNN

use CNN based on AlexNet/VGG/GoogLeNet to implement the classification of handwriting digits (not MNIST) - pytorch

使用卷积神经网络(基于AlexNet/VGG-16/GoogLeNet)实现了对手写数字数据集的识别分类

数据集为非MNIST手写数字集(由模式识别课程老师提供)共0-9十个数字,每个数字55张图片,存放在digit文件夹中,已整理规整,建议使用torchvision.datasets.ImageFolder读取数据集

图像预处理:中心裁剪+压缩

损失函数:交叉熵

优化方法:带动量的随机梯度下降

结果(测试正确率):

AlexNet:95.45%

VGG16:92.73%

GoogLeNet:97.27%

About

use CNN based on AlexNet to implement the classification of handwriting digits (not MNIST) - pytorch

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages