1. c++使用onnxruntime进行推理

link
code in git

#include <opencv2/core.hpp>
#include <opencv2/imgcodecs.hpp>
#include <opencv2/opencv.hpp>
#include <opencv2/highgui.hpp>
#include <opencv2/core/core.hpp>
#include <opencv2/imgproc/imgproc_c.h>
#include <opencv2/dnn.hpp>
#include <iostream>  
#include <onnxruntime_cxx_api.h>
#include <assert.h>
#include <vector>
#include <fstream>using namespace cv;     //当定义这一行后，cv::imread可以直接写成imread
using namespace std;
using namespace Ort;
using namespace cv::dnn;String labels_txt_file = "F:\\Pycharm\\PyCharm_Study\\Others\\c++_learning\\C++_Master\\Onnx\\classification\\classification_classes_ILSVRC2012.txt";
vector<String> readClassNames();                  // string对象作为vector对象 // 图像处理  标准化处理
void PreProcess(const Mat& image, Mat& image_blob)
{Mat input;image.copyTo(input);//数据处理 标准化std::vector<Mat> channels, channel_p;split(input, channels);Mat R, G, B;B = channels.at(0);G = channels.at(1);R = channels.at(2);B = (B / 255. - 0.406) / 0.225;G = (G / 255. - 0.456) / 0.224;R = (R / 255. - 0.485) / 0.229;channel_p.push_back(R);channel_p.push_back(G);channel_p.push_back(B);Mat outt;merge(channel_p, outt);image_blob = outt;
}// 读取txt文件
std::vector<String> readClassNames()
{std::vector<String> classNames;std::ifstream fp(labels_txt_file);if (!fp.is_open()){printf("could not open file...\n");exit(-1);}std::string name;while (!fp.eof()){std::getline(fp, name);if (name.length())classNames.push_back(name);}fp.close();return classNames;
}int main()         // 返回值为整型带参的main函数. 函数体内使用或不使用argc和argv都可
{//environment （设置为VERBOSE（ORT_LOGGING_LEVEL_VERBOSE）时，方便控制台输出时看到是使用了cpu还是gpu执行）Ort::Env env(ORT_LOGGING_LEVEL_WARNING, "OnnxModel");Ort::SessionOptions session_options;// 使用1个线程执行op,若想提升速度，增加线程数session_options.SetIntraOpNumThreads(1);CUDA加速开启(由于onnxruntime的版本太高，无cuda_provider_factory.h的头文件，加速可以使用onnxruntime V1.8的版本)//OrtSessionOptionsAppendExecutionProvider_CUDA(session_options, 0);// ORT_ENABLE_ALL: 启用所有可能的优化session_options.SetGraphOptimizationLevel(GraphOptimizationLevel::ORT_ENABLE_ALL);//load  model and creat session#ifdef _WIN32const wchar_t* model_path = L"F:\\Pycharm\\PyCharm_Study\\Others\\c++_learning\\C++_Master\\Onnx\\classification\\vgg16.onnx";
#elseconst char* model_path = "F:\\Pycharm\\PyCharm_Study\\Others\\c++_learning\\C++_Master\\Onnx\\classification\\vgg16.onnx";
#endifprintf("Using Onnxruntime C++ API\n");Ort::Session session(env, model_path, session_options);// print model input layer (node names, types, shape etc.)Ort::AllocatorWithDefaultOptions allocator;//model info// 获得模型又多少个输入和输出，一般是指对应网络层的数目// 一般输入只有图像的话input_nodes为1size_t num_input_nodes = session.GetInputCount();// 如果是多输出网络，就会是对应输出的数目size_t num_output_nodes = session.GetOutputCount();printf("Number of inputs = %zu\n", num_input_nodes);printf("Number of output = %zu\n", num_output_nodes);//获取输入nameconst char* input_name = session.GetInputName(0, allocator);               std::cout << "input_name:" << input_name << std::endl;//获取输出nameconst char* output_name = session.GetOutputName(0, allocator);std::cout << "output_name: " << output_name << std::endl;// 自动获取维度数量auto input_dims = session.GetInputTypeInfo(0).GetTensorTypeAndShapeInfo().GetShape();auto output_dims = session.GetOutputTypeInfo(0).GetTensorTypeAndShapeInfo().GetShape();std::cout << "input_dims:" << input_dims[0] << std::endl;std::cout << "output_dims:" << output_dims[0] << std::endl;    std::vector<const char*> input_names{ input_name };std::vector<const char*> output_names = { output_name };std::vector<const char*> input_node_names = { "input.1" };std::vector<const char*> output_node_names = { "70"};//加载图片Mat img = imread("F:\\Pycharm\\PyCharm_Study\\Others\\c++_learning\\C++_Master\\Onnx\\classification\\dog.jpg");Mat det1, det2;resize(img, det1, Size(256, 256), INTER_AREA);det1.convertTo(det1, CV_32FC3);PreProcess(det1, det2);         //标准化处理Mat blob = dnn::blobFromImage(det2, 1., Size(224, 224), Scalar(0, 0, 0), false, true);printf("Load success!\n");clock_t startTime, endTime;//创建输入tensorauto memory_info = Ort::MemoryInfo::CreateCpu(OrtAllocatorType::OrtArenaAllocator, OrtMemType::OrtMemTypeDefault);std::vector<Ort::Value> input_tensors;input_tensors.emplace_back(Ort::Value::CreateTensor<float>(memory_info, blob.ptr<float>(), blob.total(), input_dims.data(), input_dims.size()));/*cout << int(input_dims.size()) << endl;*/startTime = clock();推理(score model & input tensor, get back output tensor)auto output_tensors = session.Run(Ort::RunOptions{ nullptr }, input_node_names.data(), input_tensors.data(), input_names.size(), output_node_names.data(), output_node_names.size());endTime = clock();assert(output_tensors.size() == 1 && output_tensors.front().IsTensor());//除了第一个节点外，其他参数与原网络对应不上程序就会无法执行//第二个参数代表输入节点的名称集合//第四个参数1代表输入层的数目//第五个参数代表输出节点的名称集合//最后一个参数代表输出节点的数目获取输出(Get pointer to output tensor float values)float* floatarr = output_tensors[0].GetTensorMutableData<float>();     // 也可以使用output_tensors.front(); 获取list中的第一个元素变量  list.pop_front(); 删除list中的第一个位置的元素// 得到最可能分类输出Mat newarr = Mat_<double>(1, 1000); //定义一个1*1000的矩阵for (int i = 0; i < newarr.rows; i++){for (int j = 0; j < newarr.cols; j++) //矩阵列数循环{newarr.at<double>(i, j) = floatarr[j];}}/*cout << newarr.size() << endl;*/vector<String> labels = readClassNames();for (int n = 0; n < newarr.rows; n++) {Point classNumber;double classProb;Mat probMat = newarr(Rect(0, n, 1000, 1)).clone();Mat result = probMat.reshape(1, 1);minMaxLoc(result, NULL, &classProb, NULL, &classNumber);int classidx = classNumber.x;printf("\n current image classification : %s, possible : %.2f\n", labels.at(classidx).c_str(), classProb);// 显示文本putText(img, labels.at(classidx), Point(10, 20), FONT_HERSHEY_SIMPLEX, 0.6, Scalar(0, 0, 255), 1, 1);imshow("Image Classification", img);waitKey(0);}计算运行时间std::cout << "The run time is:" << (double)(endTime - startTime) / CLOCKS_PER_SEC << "s" << std::endl;printf("Done!\n");system("pause");return 0;
}