Convolution and Max Pooling of CNN (卷积和池化的实现)

最新推荐文章于 2022-05-30 10:11:56 发布

shuiliusheng

最新推荐文章于 2022-05-30 10:11:56 发布

阅读量735

点赞数

分类专栏： computer 文章标签： cnn c++ convolution and max pooling

本文链接：https://blog.csdn.net/shuiliusheng/article/details/79949113

版权

本文详细介绍了卷积神经网络（CNN）中的卷积和最大池化概念，提供了清晰直观的解释，并给出了C++实现的步骤，包括卷积模板、输入输出数据格式、卷积操作以及最大池化的原理和具体实现。还提供了示例结果和github源代码链接。

摘要由CSDN通过智能技术生成

Convolution and Max Pooling of CNN (卷积和池化的实现)

卷积和池化的具体解释（比较清晰直观）：
http://www.hackcv.com/index.php/archives/104/?hmsr=toutiao.io&utm_medium=toutiao.io&utm_source=toutiao.io

图像卷积的实现

图像卷积的原理图解
具体实现——卷积模板（Filter）

double filter0[3][3]={
    {0,0,0},{0,1,0},{0,0,0}};
double filter1[3][3]={
    {1/16.0,1/16.0,1/16.0},{1/16.0,8/16.0,1/16.0},{1/16.0,1/16.0,1/16.0}};

double filter2[3][3]={
    {-1,0,0},{0,1,0},{0,0,0}};
double filter3[3][3]={
    {0,-1,0},{0,1,0},{0,0,0}};
double filter4[3][3]={
    {0,0,0},{-1,0,1},{0,0,0}};
double filter5[3][3]={
    {1.0/9,1.0/9,1.0/9},{1.0/9,1.0/9,1.0/9},{1.0/9,1.0/9,1.0/9}};
double filter6[3][3]={
    {0,0,0},{-1,2,-1},{0,0,0}};
double filter7[3][3]={
    {0,1,0},{1,-4,1},{0,1,0}};

double filter8[3][3]={
    {0,0,-1},{0,1,0},{0,0,0}};
double filter9[3][3]={
    {0,0,0},{-1,1,0},{0,0,0}};
double filter10[3][3]={
    {0,-1,0},{0,0,0},{0,1,0}};
double filter11[3][3]={
    {1/16.0,2/16.0,1/16.0},{2/16.0,4/16.0,2/16.0},{1/16.0,2/16.0,1/16.0}};
double filter12[3][3]={
    {0,-1,0},{0,2,0},{0,-1,0}};
double filter13[3][3]={
    {1,1,1},{1,-7,1},{1,1,1}};

double filter14[3][3]={
    {0,0,0},{0,1,-1},{0,0,0}};
double filter15[3][3]={
    {0,0,0},{0,1,0},{-1,0,0}};
double filter16[3][3]={
    {-1,0,0},{0,0,0},{0,0,1}};
double filter17[3][3]={
    {1/16.0,1/16.0,1/16.0},{2/16.0,6/16.0,2/16.0},{1/16.0,1/16.0,1/16.0}};
double filter18[3][3]={
    {-1,0,0},{0,2,0},{0,0,-1}};
double filter19[3][3]={
    {-1,-1,-1},{-1,8,-1},{-1,-1,-1}};

double filter20[3][3]={
    {0,0,0},{0,1,0},{0,-1,0}};
double filter21[3][3]={
    {0,0,0},{0,1,0},{0,0,-1}};
double filter22[3][3]={
    {0,0,-1},{0,0,0},{1,0,0}};
double filter23[3][3]={
    {1/16.0,2/16.0,1/16.0},{1/16.0,6/16.0,1/16.0},{1/16.0,2/16.0,1/16.0}};
double filter24[3][3]={
    {0,0,-1},{0,2,0},{-1,0,0}};
double filter25[3][3]={
    {0,-1,0},{-