Cifar数据集转图片（jpg）

Posted 2020-10-16 jiahu

• 依赖于opencv
• 这里有已经提取好数据

#include<iostream>
#include<fstream>
#include<string>
#include<map>
#include<vector>
#include<opencv2/opencv.hpp>

using namespace std;
using namespace cv;

#define CIFAR10 //or #define CIFAR100
#ifdef CIFAR10
    const int  FRAME_SIZE = 3073;
#endif // CIFAR10
#ifdef CIFAR100
    const int  FRAME_SIZE = 3074;
#endif // CIFAR100

struct CifarImg
{
    CifarImg() : coarse_label{0}, fine_label{0}, label{-1} { }
    unsigned char coarse_label;
    unsigned char fine_label;
    char label;//just for cifar10
    Mat img;
};
//extract label info from cifar img frame
//<1 x label><<1024bytes red channel pixel><1024bytes green><1024bytes blue>>
CifarImg cifar10_img_info (const char* cifar_img_frame);
//<1 x coarse label><1 x fine label><<1024bytes red channel><1024bytes green><1024bytes blue>>
CifarImg cifar100_img_info (const char* cifar_img_frame);


int main()
{
    //使用cat data_batch_1.bin data_batch_2.bin data_batch_3.bin data_batch_4.bin data_batch_5.bin > data_train.bin 可以将cifar10中分散的数据合并，以便于处理
    const string cifar_path (R"(C:\Users\HJiahu\Desktop\cifar-10-binary.tar\cifar-10-binary\cifar-10-batches-bin\test_batch.bin)");
    const string imgs_dst_path (R"(C:\Users\HJiahu\Desktop\cifar-10-binary.tar\cifar-10-binary\cifar-10-batches-bin\test_imgs)");
    ifstream cifar_file_h (cifar_path, ifstream::binary);
    decltype (cifar100_img_info) *cifar_img_info;
#if defined(CIFAR10)
    cifar_img_info = cifar10_img_info;
#elif defined(CIFAR100)
    cifar_img_info = cifar100_img_info;
#else
    cerr << "Please define Macro CIFAR10 or CIFAR100 in src." << endl;
    getchar();
    exit (1);
#endif
    char frame_buf[FRAME_SIZE];
    int cnt = 0;
    
    while (true)
    {
        cnt = cnt + 1;
        
        if (cnt % 1000 == 0) { cout << ‘.‘; }
        
        cifar_file_h.read (frame_buf, FRAME_SIZE);
        
        //只有在尝试读取非文件内容（即文件最后一个字节之后内容）时eof()才会返回true
        //所以这里每读一次判断一次是否成功
        if (!cifar_file_h.good()) { break; }
        
        auto info = cifar_img_info (frame_buf);
        string img_name;
        
        if (info.label != -1)
        {
            img_name = to_string (info.label) + "_" + to_string (cnt) + ".jpg";
        }
        
        else
        {
            img_name = to_string (info.coarse_label) + "_"                        + to_string (info.fine_label) + "_"                        + to_string (cnt) + ".jpg";
            //cout << cnt << endl;
        }
        
        imwrite (imgs_dst_path + "\\" + img_name, info.img);
    }
}


CifarImg cifar10_img_info (const char* cifar_img_frame)
{
    CifarImg info;
    info.label = cifar_img_frame[0];
    Mat r_chl (32, 32, CV_8UC1, (char*) cifar_img_frame + 1);
    Mat g_chl (32, 32, CV_8UC1, (char*) cifar_img_frame + 1 + 1024);
    Mat b_chl (32, 32, CV_8UC1, (char*) cifar_img_frame + 1 + 1024 + 1024);
    vector<Mat> splited_mat = {b_chl, g_chl, r_chl};
    //merge (splited_mat, info.img)在windows下opencv3.1报错
    merge(&splited_mat[0], 3, info.img);
    return info;
}

CifarImg cifar100_img_info (const char* cifar_img_frame)
{
    CifarImg info;
    //info.label = cifar_img_frame[0];
    info.coarse_label = cifar_img_frame[0];
    info.fine_label = cifar_img_frame[1];
    Mat r_chl (32, 32, CV_8UC1, (char*) cifar_img_frame + 2);
    Mat g_chl (32, 32, CV_8UC1, (char*) cifar_img_frame + 2 + 1024);
    Mat b_chl (32, 32, CV_8UC1, (char*) cifar_img_frame + 2 + 1024 + 1024);
    vector<Mat> splited_mat = { b_chl, g_chl, r_chl };
    merge (&splited_mat[0], 3, info.img);
    return info;
}