矩视智能AI视觉软件-多模型单线程调用（C#）

    sdk如何加载多个模型在单线程中进行预测。确切地说，是逐个加载这些模型，并依次调用每个模型的预测接口，并将结果串行输出。

#pragma comment(lib,"Msi.lib")
#define _SILENCE_EXPERIMENTAL_FILESYSTEM_DEPRECATION_WARNING
#include <iostream>
#include <string>
#include <Windows.h>
#include <experimental/filesystem>
#include <iostream>
#include <direct.h>
#include <vector>
#include <neuro_core.h>
#include <opencv2/opencv.hpp>
#include "putTextZH.h"
#include <Windows.h>
#include <msi.h>

using namespace std;
namespace fs = experimental::filesystem;

// Get all the files's names in the dictionary, and the result will be put into the parameter vFileNames.
// 
// Parameters
// 
//         dir           [IN]           the name of dictionary.
//         vFileNames    [OUT]          the result will be put into it.
//         extension     [IN]           ZERO means having extension, such as A.jpg.
//                                      ONE means having no extension, such as A.
//  
// Returns  
//         The size of all the files.      

int getFilesName(const std::string dir, std::vector<std::string>& vFileNames, bool extension = 1)
{
    for (const auto& entry : fs::directory_iterator(dir))
    {
        if (1 == extension)
        {
            // with extension
            vFileNames.push_back(entry.path().filename().string());
        }
        else if (0 == extension)
        {
            // without extension
            std::string fileName;
            fileName = entry.path().filename().string();
            int pos = fileName.rfind(".");
            vFileNames.push_back(std::string(fileName, 0, pos));
            fileName.clear();
        }
    }
    if (vFileNames.empty()) {
        std::cout << "This dictionary is Empty, please check your dictinoary path." << endl;
    }   
     return vFileNames.size();

}

// Mask is used to cover non areas of interest, highlight areas of interest, 
// and focus image processing only on the ROI part.
// 
// Here marks the results in the picture used for segmentation.
// 
// The rectangle box x0,y0 in the result is the upper left corner,
// and the height and width of the mask are the height and width.
// 
// Parameters
//         image           [IN]           the image in Mat format.
//         DetectionResult [IN]           the result after detection.
//  
//

void drawMask(cv::Mat& image, const DetectionResult& info)
{

    cv::Mat imgMask = info.mask;
    auto x0 = std::max(std::floor(info.box.x0) - 1, 0.f);
    auto y0 = std::max(std::floor(info.box.y0) - 1, 0.f);
    cv::Rect roi((int)x0, (int)y0, info.mask_width, info.mask_height);

    // split the RGB channels, overlay mask to a specific color channel
    cv::Mat ch[3];
    split(image, ch);
    int col = 0;
    cv::bitwise_or(imgMask, ch[col](roi), ch[col](roi));   // The pixel value of each pixel is bitwise or
    merge(ch, 3, image);

}

// visualize the result by drawing the picture in the new window.
// 
// Parameters
//         image           [IN]           the image in Mat format. 
//         info            [IN]           the results after detection.
//         window_name     [IN]           the new window's name.                   
//  
// 

void visualResult(cv::Mat& image, const std::vector<DetectionResult>& info, const std::string& window_name)
{
    if (image.empty())
    {
        cerr << "input is empty, please check the path!" << std::endl;
        return;
    }

    int fontface = cv::FONT_HERSHEY_PLAIN;
    double fontscale = 1;
    int thickness = 1;

    for (int i = 0; i < info.size(); i++)
    {
        //  Generate two points and show the label on the graph.
        cv::Point p1(info[i].box.x0, info[i].box.y0), p2(info[i].box.x1, info[i].box.y1);
        putTextZH(image, (info[i].label).c_str(), p1, cv::Scalar(0, 0, 255), 50, "Arial", false, false);
        if (info[i].mask.empty()) {
            cv::rectangle(image, p1, p2, cv::Scalar(0, 255, 0), 2);
        }
        drawMask(image, info[i]);
    }

    // create a new display window and specify the type of window and then show in the screen.
    cv::namedWindow(window_name, cv::WINDOW_FREERATIO);
    cv::imshow(window_name, image);

    // wait until user presses any key to exit.
    cv::waitKey(0);

    // to close the window and de-allocate any associated memory usage.
    // For a simple program, you do not really have to call these functions because all the resources and windows of the application are closed automatically by the operating system upon exit.
    cv::destroyAllWindows();
}

int predict(string model_name, string file_path, vector<string> img_paths, int i) {

    vector<cv::Mat> images;                               // the information in opencv mat format.
    vector<int> image_ids;                                // the image's id 
    vector<cv::Mat> mats;                                 // the information in opencv mat format.

    //  predict and print the result.
    //  batch_size is given in the file model.conf. the default is ONE.

    cout << endl;
    auto img = cv::imread(file_path + "\\" + img_paths[i]);
    if (!img.data) {
        cerr << "Load image: " << img_paths[i] << "is failed" << endl;
        return -1;
    }
    else {
        cout << "Load image: " << img_paths[i] << "is successful" << endl;
    }
    images.push_back(img);
    image_ids.push_back(i);
    mats.push_back(img);
    cout << img_paths[i] << endl;
    if ((int)mats.size() == get_batch(model_name.c_str())) {
        vector<vector<DetectionResult>> out_results{};
        DWORD start = GetTickCount64();        // beginning time
        int predictStatus = predict_model(model_name.c_str(), mats, out_results);
        DWORD end = GetTickCount64();          //  end time
        if (predictStatus != 0) {
            cout << "This prediction is failed " << endl;
            cout << "Time cost:                          " << end - start << " ms" << endl << endl << endl;
            return -1;
        }
        // The results to be printed.
        // 
        // for OCR and object Detection, results are rectangle box, confidence level,category.
        // 
        // for Pixel Segmentation, results are rectangle box, confidence level,category, pixel segmentation image.
        int index = 0;
        for (auto res : out_results) {
            for (auto r : res) {
                cout << endl;
                cout << "label:                              " << r.label << endl;
                cout << "Label_index:                        " << r.label_index << endl;
                cout << "Confidential score:                 " << r.score << endl;
                cout << "Position of result(x0,y0,x1,y1):    " << r.box.x0 << "  " << r.box.y0 << "  " << r.box.x1 << "  " << r.box.y1 << endl;
                cout << "Row_index:                          " << r.row_index << endl;
                cout << "Col_index:                          " << r.col_index << endl;
                cout << "Mask_width:                         " << r.mask_width << endl;
                cout << "Mask_height:                        " << r.mask_height << endl;
            }
            cout << endl;
            cout << "Time cost:                          " << end - start << " ms" << endl << endl << endl;
            cout << "Press enter to continue             " << endl;
            // It will visual a result by a new window whose name is show.
            visualResult(mats[index], res, img_paths[index]);
            index++;
        }
        mats.clear();
        image_ids.clear();
        images.clear();
        cout << endl << endl;
    }
    return 1;
}

int main(int argc, char ** argv) {

    
    //  model_path is the path of the model,    such as "C:\\Users\\NeuroBot\\A"
    //  file_path  is the path of the pictures, such as "C:\\Users\\NeuroBot\\picture"
    string device_name = "cuda";
    string model_path_1 = "C:\\Users\\NeuroBot\\A";
    string model_path_2 = "C:\\Users\\NeuroBot\\B";
    string file_path_1 = "C:\\Users\\NeuroBot\\pictureA";
    string file_path_2 = "C:\\Users\\NeuroBot\\pictureB";
    string model_name_1 = "A";
    string model_name_2 = "B";
    int status = load_model(model_name_1.c_str(), model_path_1.c_str(), device_name.c_str());
    if (status != 0) {
        cerr << "failed to create detector, code: " << status << endl;
        return -1;
    }
    status = load_model(model_name_2.c_str(), model_path_2.c_str(), device_name.c_str());
    if (status != 0) {
        cerr << "failed to create detector, code: " << status << endl;
        return -1;
    }
    vector<string> img_paths_1, img_paths_2;                  // the list of the pictures' name 
    getFilesName(file_path_1, img_paths_1);                   // get all the files's names in the dictionary,
    getFilesName(file_path_2, img_paths_2);                   // get all the files's names in the dictionary,
    int i = 0, j = 0, count = 0;
    while(i < img_paths_1.size() && j < img_paths_2.size()){
        if(count % 2 == 0){
            predict(model_name_1, file_path_1, img_paths_1, i);
            ++i;
        }
        else{
            predict(model_name_2, file_path_2, img_paths_2, j);
            ++j;
        }
        count++;
    }

    while(i < img_paths_1.size()){
        predict(model_name_1, file_path_1, img_paths_1, i);
        ++i;
    }
    while(j < img_paths_2.size()){
        predict(model_name_2, file_path_2, img_paths_2, j);
        ++j;

    }
    return 0;

}

附录

1：加载错误码定义：

地位	定义
0	成功
1	未找到配置文件
2	未找到批处理参数
3	未找到dete_thres参数
4	未找到类参数
5	缺少执行所需的 DLL
6	ARG - NVIDIA 显卡错误
7	不支持
8	出界
9	内存不足
10	缺少用于加载的模型相关文件
11	加载失败
12	状态计数
50	模型类型错误或无法读取 model.conf 文件

2：预测错误码定义：

地位	定义
0	成功
1	未找到配置文件
2	未找到批处理参数
3	未找到dete_thres参数
4	未找到类参数
5	缺少执行所需的 DLL
6	ARG - NVIDIA 显卡错误
7	不支持
50	授权文件不存在或读取权限被拒绝
51	内存不足
52	缺少用于加载的模型相关文件
53	加载失败
54	状态计数
55	模型类型错误或无法读取 model.conf 文件
56	授权文件没有写入权限
57	与授权相关的未知错误
58	授权文件版本与 SDK 版本不同
59	超出标签数量限制