Skip to content
/ dataIO Public

A header-only library for convenient I/O in computer vision.

License

Notifications You must be signed in to change notification settings

yipingp/dataIO

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

18 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

dataIO

Introduction

A header only library for convenient coding in the field of computer vision.

Supported features

  • Data (including txt, OpenCV image, PCL point cloud) input and output

  • Get files in a folder

  • Read/Write RGBD dataset

  • TCP/IP

  • Point cloud rendering.

Dependencies

  • Linux (tested on Ubuntu 18.04)
  • OpenCV
  • PCL (only required when PointCloudIO.hpp is included)

Usage

Read files from a folder

void dataIO::getAllFiles(string suffix = "");

Complete version:

#include "dataIO.hpp"

int main()
{
    dataIO files;
    files.path=".";
    files.getAllFiles(".txt"); // Read files with specified suffix

    // filename
    for (auto ownname:files.ownnames)
        cout << ownname << endl;

    // directory + filename
    for (auto filepath : files.filepaths)
        cout << filepath << endl;
}

Read RGB-D dataset

Reading dataset downloaded from here or of your own (detailed below) is both available.

template <typename T>
T dataIO::getImage<T>();
T dataIO::getDepth<T>();

Complete version:

#include "dataIO.hpp"

/************Parameters************/
string RGBD_dataset_folder_dir = "./RGBD_dataset";

int main()
{
    dataIO colorIO, depthIO;
    colorIO.path = RGBD_dataset_folder_dir;
    colorIO.filename = "rgb.txt";
    depthIO.path = colorIO.path;
    depthIO.filename = "depth.txt";
    
    while(true)
    {
        cv::Mat color = colorIO.getImage<Mat>();
        cv::Mat depth = depthIO.getDepth<Mat>();
    }
}

Write RGB-D dataset

For users of RGB-D cameras, we create a dataset with the following structure under dataset_dir.

RGBD_dataset

├── depth

│ ├── 1.2.png

│ ├── 2.2.png

│ └── ...

├── depth.txt

├── rgb

│ ├── 1.1.png

│ ├── 2.1.png

│ └── ...

└── rgb.txt

#include "dataIO.hpp"

/************Parameters************/
string dataset_dir = ".";
string dataset_folder = "RGBD_dataset";

int main()
{
    dataIO files;
    files.path = dataset_dir;
    vector<string> colorNames, depthNames; // Names of color images and depth images

    for (int i = 0; i < 10; i++)
    {
        files.save_RGBD_dataset(dataset_folder, colorNames, depthNames); // First we create a folder named RGBD_dataset under files.path

        static int num_cnt = 1;
        stringstream file_cnt;
        file_cnt << num_cnt;
        num_cnt++;
        // ".1" is used to name color images, the first color image is named "1.1.png"
        // ".2" is used to name depth images, the first color image is named "1.2.png"
        string color_path = files.joinPath(files.path, dataset_folder) + "/" + files.joinPath("rgb", file_cnt.str()) + ".1.png";
        string depth_path = files.joinPath(files.path, dataset_folder) + "/" + files.joinPath("depth", file_cnt.str()) + ".2.png";
        colorNames.push_back(color_path);
        depthNames.push_back(depth_path);

        cv::Mat color;
        cv::Mat depth;
        /*
        image caputring..
        complemented by your own depending on your camera
        */
        cv::imwrite(color_path, color); // Write captured color image and depth image
        cv::imwrite(depth_path, depth);
    }
    files.save_RGBD_dataset("RGBD_dataset", colorNames, depthNames); // write rgb.txt and depth.txt
}

TCP/IP Server

#include "dataIO.hpp"

int main()
{
    dataIO files;
    files.tcpipInit(6666); // SERVER_PORT is 6666
    
    // After connected
    string str = "message";
    files.tcpipSend(str); // Send message to client
}

IO and visualization of point clouds

template <class PointType>
void PointCloudIO::PCRender(pcl::PointCloud<PointType> cloudView1, pcl::PointCloud<PointType> cloudView2 = pcl::PointCloud<PointType>(), pcl::PointCloud<PointType> cloudView3 = pcl::PointCloud<PointType>(), pcl::PointCloud<PointType> cloudView4 = pcl::PointCloud<PointType>())

Complete version:

#included "PointCloudIO.hpp"

typedef pcl::PointXYZRGB PointT;

int main()
{
	PointCloudIO<PointT> file;
	file.path = "cloud_files";

	file.filename = "cloud1.ply"; pcl::PointCloud<PointT> cloud1 = file.readPC();
	file.filename = "cloud2.ply"; pcl::PointCloud<PointT> cloud2 = file.readPC();

	// optinal settings for PC rendering
	/*
	file.set_txt_content("cloud1", "cloud2"); // set text content respectively
	file.set_txt_rgb(pcl::PointXYZ(0.7, 0.7, 0.7), pcl::PointXYZ(0.7, 0.7, 0.7)); // set text color respectively
	file.rgb_bkg = pcl::PointXYZ(0.3, 0.3, 0.3); // set backgound color
	file.coordinateScale = 0.05; // set scale of coordinate
	*/

	file.PCrender(cloud1, cloud2); // 4 point clouds for visualization at most

	return 0;
}

About

A header-only library for convenient I/O in computer vision.

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published