convert big yml data

Question

There is any other way/trick to store the depth map in the database? Basically, we are trying to store 300000 double values. If it helps we can convert the array into a NSMutabl

Answer 1

You can save a lot of memory by storing the raw binary data in a BLOB.

If you don't have fixed rows and cols of your matrix, you can put at the beginning of the file two integers for rows and cols.

I'll add a simple example on how to save and load the data of matrix, preceded by rows and cols.

#include <opencv2/opencv.hpp>
#include <fstream>

using namespace cv;
using namespace std;

int main()
{
    Mat1d m = (Mat1d(3, 3) << 1, 2, 3, 4, 5, 6, 7, 8, 9);
    Mat1d n;

    {
        ofstream ofs("test.bin", fstream::binary);
        ofs.write((char*)&m.rows, sizeof(int)); // Save rows
        ofs.write((char*)&m.cols, sizeof(int)); // Save cols
        ofs.write((char*)m.data, m.total()*sizeof(double)); // Save data
    }

    {
        ifstream ifs("test.bin", fstream::binary);
        int rows, cols;
        ifs.read((char*)&rows, sizeof(int)); // Load rows
        ifs.read((char*)&cols, sizeof(int)); // Load cols

        n = Mat1d(rows, cols); // Resize the matrix according to rows, cols
        ifs.read((char*)n.data, rows*cols*sizeof(double)); // Load data
    }

    // Now m and n are equal
    return 0;
}

If you need further compression you can read and write the stream using gzstream