first init

src/yolov8Predictor.cpp

+#include "yolov8Predictor.h"
+
+YOLOPredictor::YOLOPredictor(const std::string &modelPath,
+                             const bool &isGPU,
+                             float confThreshold,
+                             float iouThreshold,
+                             float maskThreshold)
+{
+    this->confThreshold = confThreshold;
+    this->iouThreshold = iouThreshold;
+    this->maskThreshold = maskThreshold;
+    env = Ort::Env(OrtLoggingLevel::ORT_LOGGING_LEVEL_WARNING, "YOLOV8");
+    sessionOptions = Ort::SessionOptions();
+
+    // std::vector<std::string> availableProviders = Ort::GetAvailableProviders();
+    // auto cudaAvailable = std::find(availableProviders.begin(), availableProviders.end(), "CUDAExecutionProvider");
+    // OrtCUDAProviderOptions cudaOption;
+    OrtROCMProviderOptions rocmOption;
+    rocmOption.device_id = 0;
+
+    std::cout << "Inference device: DCU" << std::endl;
+    // sessionOptions.AppendExecutionProvider_CUDA(cudaOption);
+    sessionOptions.AppendExecutionProvider_ROCM(rocmOption);
+
+    session = Ort::Session(env, modelPath.c_str(), sessionOptions);
+
+    const size_t num_input_nodes = session.GetInputCount();   //==1
+    const size_t num_output_nodes = session.GetOutputCount(); //==1,2
+    if (num_output_nodes > 1)
+    {
+        this->hasMask = true;
+        std::cout << "Instance Segmentation" << std::endl;
+    }
+    else
+        std::cout << "Object Detection" << std::endl;
+
+    Ort::AllocatorWithDefaultOptions allocator;
+    for (int i = 0; i < num_input_nodes; i++)
+    {
+        auto input_name = session.GetInputNameAllocated(i, allocator);
+        this->inputNames.push_back(input_name.get());
+        input_names_ptr.push_back(std::move(input_name));
+
+        Ort::TypeInfo inputTypeInfo = session.GetInputTypeInfo(i);
+        std::vector<int64_t> inputTensorShape = inputTypeInfo.GetTensorTypeAndShapeInfo().GetShape();
+        this->inputShapes.push_back(inputTensorShape);
+        this->isDynamicInputShape = false;
+        // checking if width and height are dynamic
+        if (inputTensorShape[2] == -1 && inputTensorShape[3] == -1)
+        {
+            std::cout << "Dynamic input shape" << std::endl;
+            this->isDynamicInputShape = true;
+        }
+    }
+    for (int i = 0; i < num_output_nodes; i++)
+    {
+        auto output_name = session.GetOutputNameAllocated(i, allocator);
+        this->outputNames.push_back(output_name.get());
+        output_names_ptr.push_back(std::move(output_name));
+
+        Ort::TypeInfo outputTypeInfo = session.GetOutputTypeInfo(i);
+        std::vector<int64_t> outputTensorShape = outputTypeInfo.GetTensorTypeAndShapeInfo().GetShape();
+        this->outputShapes.push_back(outputTensorShape);
+        if (i == 0)
+        {
+            if (!this->hasMask)
+                classNums = outputTensorShape[1] - 4;
+            else
+                classNums = outputTensorShape[1] - 4 - 32;
+        }
+    }
+}
+
+void YOLOPredictor::getBestClassInfo(std::vector<float>::iterator it,
+                                     float &bestConf,
+                                     int &bestClassId,
+                                     const int _classNums)
+{
+    // first 4 element are box
+    bestClassId = 4;
+    bestConf = 0;
+
+    for (int i = 4; i < _classNums + 4; i++)
+    {
+        if (it[i] > bestConf)
+        {
+            bestConf = it[i];
+            bestClassId = i - 4;
+        }
+    }
+}
+
+
+cv::Mat YOLOPredictor::getMask(const cv::Mat &maskProposals,
+                               const cv::Mat &maskProtos)
+{
+    cv::Mat protos = maskProtos.reshape(0, {(int)this->outputShapes[1][1], (int)this->outputShapes[1][2] * (int)this->outputShapes[1][3]});
+
+    cv::Mat matmul_res = (maskProposals * protos).t();
+    cv::Mat masks = matmul_res.reshape(1, {(int)this->outputShapes[1][2], (int)this->outputShapes[1][3]});
+    cv::Mat dest;
+
+    // sigmoid
+    cv::exp(-masks, dest);
+    dest = 1.0 / (1.0 + dest);
+    cv::resize(dest, dest, cv::Size((int)this->inputShapes[0][2], (int)this->inputShapes[0][3]), cv::INTER_LINEAR);
+    return dest;
+}
+
+
+void YOLOPredictor::preprocessing(std::vector<cv::Mat> &images, float *&blob, std::vector<int64_t> &inputTensorShape)
+{
+    int batchSize = images.size();
+    if(batchSize == 0) return;
+    int targetWidth = (int)this->inputShapes[0][2];
+    int targetHeight = (int)this->inputShapes[0][3];
+
+    size_t totalElements = batchSize * 3 * targetWidth * targetHeight;
+    blob = new float[totalElements];
+    
+    inputTensorShape = {batchSize, 3, targetHeight, targetWidth};
+
+    for (int b = 0; b < batchSize; ++b)
+    {
+        cv::Mat resizedImage, floatImage;
+        cv::cvtColor(images[b], resizedImage, cv::COLOR_BGR2RGB);
+        utils::letterbox(resizedImage, resizedImage, cv::Size((int)this->inputShapes[0][2], (int)this->inputShapes[0][3]),
+                         cv::Scalar(114, 114, 114), this->isDynamicInputShape,
+                         false, true, 32);
+        resizedImage.convertTo(floatImage, CV_32FC3, 1 / 255.0);
+        cv::Size floatImageSize{floatImage.cols, floatImage.rows};
+
+        // hwc -> chw
+        std::vector<cv::Mat> chw(floatImage.channels());
+        for (int i = 0; i < floatImage.channels(); ++i)
+        {
+            chw[i] = cv::Mat(floatImageSize, CV_32FC1, blob +  b * 3 * floatImageSize.width * floatImageSize.height  + i * floatImageSize.width * floatImageSize.height);
+        }
+        cv::split(floatImage, chw);
+    }
+}
+
+std::vector<std::vector<Yolov8Result>> YOLOPredictor::postprocessing(const cv::Size &resizedImageShape,
+                                                                    const std::vector<cv::Size> &originalImageShapes,
+                                                                    std::vector<Ort::Value> &outputTensors)
+{
+    int batchsize = originalImageShapes.size();
+    if(batchsize == 0) return {};
+    // for box
+    std::vector<std::vector<cv::Rect>> boxes(batchsize);
+    std::vector<std::vector<float>> confs(batchsize);
+    std::vector<std::vector<int>> classIds(batchsize);
+
+    float *boxOutput = outputTensors[0].GetTensorMutableData<float>();
+    int rows = (int)this->outputShapes[0][2];
+    int cols = (int)this->outputShapes[0][1];
+    // if hasMask
+    std::vector<std::vector<std::vector<float>>> picked_proposals(batchsize);
+    std::vector<cv::Mat> mask_protos(batchsize);
+    //[1,4+n,8400]=>[1,8400,4+n] or [1,4+n+32,8400]=>[1,8400,4+n+32]
+    cv::Mat output0 = cv::Mat(cv::Size((int)this->outputShapes[0][2], (int)this->outputShapes[0][1]), CV_32F, boxOutput).t();
+    float *output0ptr = (float *)output0.data;
+    // std::cout << rows << cols << std::endl;
+    
+    for (int b = 0; b < batchsize; ++b)
+    {
+        //[1,4+n,8400]=>[1,8400,4+n] or [1,4+n+32,8400]=>[1,8400,4+n+32]
+        cv::Mat output0 = cv::Mat(cv::Size((int)this->outputShapes[0][2], (int)this->outputShapes[0][1]), CV_32F, boxOutput + b * rows * cols).t();
+        float *output0ptr = (float *)output0.data;
+        for (int i = 0; i < rows; i++)
+        {
+            std::vector<float> it(output0ptr + i * cols, output0ptr + (i + 1) * cols);
+            float confidence;
+            int classId;
+            this->getBestClassInfo(it.begin(), confidence, classId, classNums);
+
+            if (confidence > this->confThreshold)
+            {
+                if (this->hasMask)
+                {
+                    std::vector<float> temp(it.begin() + 4 + classNums, it.end());
+                    picked_proposals[b].push_back(temp);
+                }
+                int centerX = (int)(it[0]);
+                int centerY = (int)(it[1]);
+                int width = (int)(it[2]);
+                int height = (int)(it[3]);
+                int left = centerX - width / 2;
+                int top = centerY - height / 2;
+                boxes[b].emplace_back(left, top, width, height);
+                confs[b].emplace_back(confidence);
+                classIds[b].emplace_back(classId);
+            }
+        }
+        if (this->hasMask)
+        {
+            float* maskOutput = outputTensors[1].GetTensorMutableData<float>() + b * this->outputShapes[1][1] * this->outputShapes[1][2] * this->outputShapes[1][3];
+            std::vector<int> mask_protos_shape = {1, (int)this->outputShapes[1][1], (int)this->outputShapes[1][2], (int)this->outputShapes[1][3]};
+            mask_protos[b] = cv::Mat(mask_protos_shape, CV_32F, maskOutput);
+        }
+    }
+
+    std::vector<std::vector<Yolov8Result>> results(batchsize);
+    for (int b = 0; b < batchsize; ++b) {
+        std::vector<int> indices;
+        cv::dnn::NMSBoxes(boxes[b], confs[b], this->confThreshold, this->iouThreshold, indices);
+
+        for (int idx : indices) {
+            Yolov8Result res;
+            res.box = cv::Rect(boxes[b][idx]);
+            if (this->hasMask)
+                res.boxMask = this->getMask(cv::Mat(picked_proposals[b][idx]).t(), mask_protos[b]);
+            else
+                res.boxMask = cv::Mat::zeros((int)this->inputShapes[0][2], (int)this->inputShapes[0][3], CV_8U);
+
+            utils::scaleCoords(res.box, res.boxMask, this->maskThreshold, resizedImageShape, originalImageShapes[b]);
+            res.conf = confs[b][idx];
+            res.classId = classIds[b][idx];
+            results[b].emplace_back(res);
+        }
+    }
+
+    return results;
+}
+
+
+std::vector<std::vector<Yolov8Result>> YOLOPredictor::predict(std::vector<cv::Mat>& images) {
+    if (images.empty()) return {};
+
+    float* blob = nullptr;
+    std::vector<int64_t> inputTensorShape{1, 3, -1, -1};
+    this->preprocessing(images, blob, inputTensorShape);
+
+    size_t inputTensorSize = utils::vectorProduct(inputTensorShape);
+
+    std::vector<float> inputTensorValues(blob, blob + inputTensorSize);
+
+    std::vector<Ort::Value> inputTensors;
+
+    Ort::MemoryInfo memoryInfo = Ort::MemoryInfo::CreateCpu(OrtAllocatorType::OrtArenaAllocator, OrtMemType::OrtMemTypeDefault);
+
+    inputTensors.push_back(Ort::Value::CreateTensor<float>(
+        memoryInfo, inputTensorValues.data(), inputTensorSize,
+        inputTensorShape.data(), inputTensorShape.size()));
+
+    std::vector<Ort::Value> outputTensors = this->session.Run(Ort::RunOptions{nullptr},
+                                                            this->inputNames.data(),
+                                                            inputTensors.data(),
+                                                            1,
+                                                            this->outputNames.data(),
+                                                            this->outputNames.size());
+
+    cv::Size resizedShapes(cv::Size((int)inputTensorShape[3], (int)inputTensorShape[2]));
+    std::vector<cv::Size> originalShapes;
+    for (const auto& image : images) {
+        originalShapes.push_back(image.size());
+    }
+
+    std::vector<std::vector<Yolov8Result>> results = this->postprocessing(resizedShapes,
+                                                                            originalShapes,
+                                                                            outputTensors);
+
+    delete[] blob;
+    return results;
+}