ANSCORE/modules/ANSODEngine/ANSOPENVINOCL.cpp

#include "ANSOPENVINOCL.h"
#include "Utility.h"
namespace ANSCENTER
{
	bool OPENVINOCL::OptimizeModel(bool fp16, std::string& optimizedModelFolder) {
		std::lock_guard<std::recursive_mutex> lock(_mutex);
		if (!ANSODBase::OptimizeModel(fp16, optimizedModelFolder)) {
			return false;
		}
		if (FileExist(_modelFilePath)) {
			std::string modelName = GetFileNameWithoutExtension(_modelFilePath);
			std::string binaryModelName = modelName + ".bin";
			std::string modelFolder = GetParentFolder(_modelFilePath);
			std::string optimizedModelPath = CreateFilePath(modelFolder, binaryModelName);
			if (FileExist(optimizedModelPath)) {
				this->_logger.LogDebug("OPENVINOCL::OptimizeModel", "This model is optimized. No need other optimization.", __FILE__, __LINE__);
				optimizedModelFolder = modelFolder;
				return true;
			}
			else {
				this->_logger.LogFatal("OPENVINOCL::OptimizeModel", "This model can not be optimized.", __FILE__, __LINE__);
				optimizedModelFolder = modelFolder;
				return false;
			}
		}
		else {
			this->_logger.LogFatal("OPENVINOCL::OptimizeModel", "This model is not exist. Please check the model path again.", __FILE__, __LINE__);
			optimizedModelFolder = "";
			return false;
		}
	}
	bool OPENVINOCL::LoadModel(const std::string& modelZipFilePath, const std::string& modelZipPassword) {
		std::lock_guard<std::recursive_mutex> lock(_mutex);
		ModelLoadingGuard mlg(_modelLoading);
		try {
			bool result = ANSODBase::LoadModel(modelZipFilePath, modelZipPassword);
			if (!result) return false;
			// 0. Check if the configuration file exist
			if (FileExist(_modelConfigFile)) {
				ModelType modelType;
				std::vector<int> inputShape;
				_classes = ANSUtilityHelper::GetConfigFileContent(_modelConfigFile, modelType, inputShape);
				if (inputShape.size() == 2) {
					if (inputShape[0] > 0)_modelConfig.inpHeight = inputShape[0];
					if (inputShape[1] > 0)_modelConfig.inpWidth = inputShape[1];
				}
			}
			else {// This is old version of model zip file
				std::string onnxfile = CreateFilePath(_modelFolder, "train_last.xml");//yolov8n.xml
				if (std::filesystem::exists(onnxfile)) {
					_modelFilePath = onnxfile;
					_classFilePath = CreateFilePath(_modelFolder, "classes.names");
					this->_logger.LogDebug("OPENVINOCL::Initialize.  Loading OpenVINO weight", _modelFilePath, __FILE__, __LINE__);
				}
				else {
					this->_logger.LogError("OPENVINOCL::Initialize.  Model file is not exist", _modelFilePath, __FILE__, __LINE__);
					return false;
				}
				std::ifstream isValidFileName(_classFilePath);
				if (!isValidFileName)
				{
					this->_logger.LogDebug("OPENVINOCL::Initialize.  Load classes from string", _classFilePath, __FILE__, __LINE__);
					LoadClassesFromString();
				}
				else {
					this->_logger.LogDebug("OPENVINOCL::Initialize.  Load classes from file", _classFilePath, __FILE__, __LINE__);
					LoadClassesFromFile();
				}
			}
			// Load Model from Here
			InitialModel();
			_isInitialized = true;
			return true;
		}
		catch (std::exception& e) {
			this->_logger.LogFatal("OPENVINOCL::LoadModel", e.what(), __FILE__, __LINE__);
			return false;
		}
	}
	bool OPENVINOCL::LoadModelFromFolder(std::string licenseKey, ModelConfig modelConfig, std::string modelName, std::string className, const std::string& modelFolder, std::string& labelMap) {
		std::lock_guard<std::recursive_mutex> lock(_mutex);
		ModelLoadingGuard mlg(_modelLoading);
		try {
			bool result = ANSODBase::LoadModelFromFolder(licenseKey, modelConfig, modelName, className, modelFolder, labelMap);
			if (!result) return false;
			std::string _modelName = modelName;
			if (_modelName.empty()) {
				_modelName = "train_last";
			}
			std::string modelFullName = _modelName + ".xml";
			// Parsing for YOLO only here
			_modelConfig = modelConfig;
			_modelConfig.detectionType = ANSCENTER::DetectionType::CLASSIFICATION;
			_modelConfig.modelType = ModelType::OPENVINO;
			_modelConfig.inpHeight = 224;
			_modelConfig.inpWidth = 224;
			if (_modelConfig.modelMNSThreshold < 0.2)
				_modelConfig.modelMNSThreshold = 0.5;
			if (_modelConfig.modelConfThreshold < 0.2)
				_modelConfig.modelConfThreshold = 0.5;
			// 0. Check if the configuration file exist
			if (FileExist(_modelConfigFile)) {
				ModelType modelType;
				std::vector<int> inputShape;
				_classes = ANSUtilityHelper::GetConfigFileContent(_modelConfigFile, modelType, inputShape);
				if (inputShape.size() == 2) {
					if (inputShape[0] > 0)_modelConfig.inpHeight = inputShape[0];
					if (inputShape[1] > 0)_modelConfig.inpWidth = inputShape[1];
				}
			}
			else {// This is old version of model zip file
				std::string onnxfile = CreateFilePath(_modelFolder, modelFullName);//yolov8n.xml
				if (std::filesystem::exists(onnxfile)) {
					_modelFilePath = onnxfile;
					_classFilePath = CreateFilePath(_modelFolder, className);
					this->_logger.LogDebug("OPENVINOCL::Initialize.  Loading OpenVINO weight", _modelFilePath, __FILE__, __LINE__);
				}
				else {
					this->_logger.LogError("OPENVINOCL::Initialize.  Model file is not exist", _modelFilePath, __FILE__, __LINE__);
					return false;
				}
				std::ifstream isValidFileName(_classFilePath);
				if (!isValidFileName)
				{
					this->_logger.LogDebug("OPENVINOCL::Initialize.  Load classes from string", _classFilePath, __FILE__, __LINE__);
					LoadClassesFromString();
				}
				else {
					this->_logger.LogDebug("OPENVINOCL::Initialize.  Load classes from file", _classFilePath, __FILE__, __LINE__);
					LoadClassesFromFile();
				}
			}

			// 1. Load labelMap and engine
			labelMap.clear();
			if (!_classes.empty())
				labelMap = VectorToCommaSeparatedString(_classes);

			// Load Model from Here
			InitialModel();
			_isInitialized = true;
			return true;
		}
		catch (std::exception& e) {
			this->_logger.LogFatal("OPENVINOCL::LoadModel", e.what(), __FILE__, __LINE__);
			return false;
		}
	}
	cv::Mat OPENVINOCL::PreProcessing(const cv::Mat& source) {
		if (source.empty()) {
			std::cerr << "Error: Empty image provided to PreProcessing" << std::endl;
			return cv::Mat();
		}

		// Convert grayscale to 3-channel BGR
		cv::Mat processedImage;
		if (source.channels() == 1) {
			cv::cvtColor(source, processedImage, cv::COLOR_GRAY2BGR);
		}
		else {
			processedImage = source;
		}

		int col = processedImage.cols;
		int row = processedImage.rows;
		int maxSize = std::max(col, row);

		// Create a square padded image with a black background
		cv::Mat result = cv::Mat::zeros(maxSize, maxSize, CV_8UC3);

		// Copy the original image to the top-left corner of the square matrix
		processedImage.copyTo(result(cv::Rect(0, 0, col, row)));

		return result;
	}
	bool OPENVINOCL::Initialize(std::string licenseKey, ModelConfig modelConfig, const std::string& modelZipFilePath, const std::string& modelZipPassword, std::string& labelMap) {
		std::lock_guard<std::recursive_mutex> lock(_mutex);
		ModelLoadingGuard mlg(_modelLoading);
		try {
			std::string openVINOVersion = ov::get_openvino_version().buildNumber;
			//this->_logger.LogDebug("OPENVINOCL::Initialize. OpenVINO version", openVINOVersion, __FILE__, __LINE__);
			bool result = ANSODBase::Initialize(licenseKey, modelConfig, modelZipFilePath, modelZipPassword, labelMap);
			if (!result) return false;
			// Parsing for YOLO only here
			_modelConfig = modelConfig;
			_modelConfig.detectionType = ANSCENTER::DetectionType::CLASSIFICATION;
			_modelConfig.modelType = ModelType::OPENVINO;
			_modelConfig.inpHeight = 224;
			_modelConfig.inpWidth = 224;
			if (_modelConfig.modelMNSThreshold < 0.2)
				_modelConfig.modelMNSThreshold = 0.5;
			if (_modelConfig.modelConfThreshold < 0.2)
				_modelConfig.modelConfThreshold = 0.5;
			// 0. Check if the configuration file exist
			if (FileExist(_modelConfigFile)) {
				ModelType modelType;
				std::vector<int> inputShape;
				_classes = ANSUtilityHelper::GetConfigFileContent(_modelConfigFile, modelType, inputShape);
				if (inputShape.size() == 2) {
					if (inputShape[0] > 0)_modelConfig.inpHeight = inputShape[0];
					if (inputShape[1] > 0)_modelConfig.inpWidth = inputShape[1];
				}
			}
			else {// This is old version of model zip file
				std::string onnxfile = CreateFilePath(_modelFolder, "train_last.xml");//yolov8n.xml
				if (std::filesystem::exists(onnxfile)) {
					_modelFilePath = onnxfile;
					_classFilePath = CreateFilePath(_modelFolder, "classes.names");
					this->_logger.LogDebug("OPENVINOCL::Initialize.  Loading OpenVINO weight", _modelFilePath, __FILE__, __LINE__);
				}
				else {
					this->_logger.LogError("OPENVINOCL::Initialize.  Model file is not exist", _modelFilePath, __FILE__, __LINE__);
					return false;
				}
				std::ifstream isValidFileName(_classFilePath);
				if (!isValidFileName)
				{
					this->_logger.LogDebug("OPENVINOCL::Initialize.  Load classes from string", _classFilePath, __FILE__, __LINE__);
					LoadClassesFromString();
				}
				else {
					this->_logger.LogDebug("OPENVINOCL::Initialize.  Load classes from file", _classFilePath, __FILE__, __LINE__);
					LoadClassesFromFile();
				}
			}

			// 1. Load labelMap and engine
			labelMap.clear();
			if (!_classes.empty())
				labelMap = VectorToCommaSeparatedString(_classes);

			// Load Model from Here
			InitialModel();
			_isInitialized = true;
			return true;

		}
		catch (std::exception& e) {
			this->_logger.LogFatal("OPENVINOCL::Initialize", e.what(), __FILE__, __LINE__);
			return false;
		}
	}
	std::vector<Object> OPENVINOCL::RunInference(const cv::Mat& input) {
		return RunInference(input, "CustomCam");
	}
	std::vector<Object> OPENVINOCL::RunInference(const cv::Mat& input,const std::string& camera_id)
	{
		if (!PreInferenceCheck("OPENVINOCL::RunInference")) return {};

		if (input.empty() || input.cols < 20 || input.rows < 20) {
			_logger.LogError("OPENVINOCL::RunInference", "Input image is invalid",
				__FILE__, __LINE__);
			return {};
		}

		try {
			// Step 1: Resize directly to target size (no clone needed!)
			cv::Mat resized;
			cv::resize(input, resized, cv::Size(224, 224), 0, 0, cv::INTER_LINEAR);

			// Step 2: Preprocessing
			cv::Mat letterbox_img = PreProcessing(resized);

			// Step 3: Create blob
			cv::Mat blob = cv::dnn::blobFromImage(letterbox_img, 1.0 / 255.0,
				cv::Size(224, 224), cv::Scalar(), true);

			// Step 4: Set input tensor
			auto input_port = compiled_model_.input();
			ov::Tensor input_tensor(input_port.get_element_type(),
				input_port.get_shape(), blob.ptr(0));
			inference_request_.set_input_tensor(input_tensor);

			// Step 5: Run inference
			inference_request_.start_async();
			inference_request_.wait();

			// Step 6: Get output
			auto output = inference_request_.get_output_tensor(0);
			auto output_shape = output.get_shape();

			if (output_shape.size() < 2) {
				_logger.LogError("OPENVINOCL::RunInference", "Invalid output shape",
					__FILE__, __LINE__);
				return {};
			}

			// Step 7: Find max class (no vector copy needed!)
			const float* output_buffer = output.data<const float>();
			const size_t num_classes = output_shape[1];

			auto max_ptr = std::max_element(output_buffer, output_buffer + num_classes);
			const int class_id = static_cast<int>(std::distance(output_buffer, max_ptr));
			const float score = *max_ptr;

			// Step 8: Create result object
			Object clsResult;
			clsResult.classId = class_id;
			clsResult.confidence = score;
			clsResult.cameraId = camera_id;

			// Set class name
			const int classNameSize = static_cast<int>(_classes.size());
			if (!_classes.empty()) {
				clsResult.className = (class_id < classNameSize)
					? _classes[class_id]
					: _classes.back();
			}
			else {
				clsResult.className = "Unknown";
			}

			// Set bounding box (full image for classification)
			clsResult.box = cv::Rect(0, 0, input.cols, input.rows);

			// Set polygon
			clsResult.polygon = ANSUtilityHelper::RectToNormalizedPolygon(
				clsResult.box, input.cols, input.rows
			);

			// Return single-element vector
			return { std::move(clsResult) };

		}
		catch (const std::exception& e) {
			_logger.LogFatal("OPENVINOCL::RunInference", e.what(), __FILE__, __LINE__);
			return {};
		}
		catch (...) {
			_logger.LogFatal("OPENVINOCL::RunInference", "Unknown exception",
				__FILE__, __LINE__);
			return {};
		}
	}
	OPENVINOCL::~OPENVINOCL() {
		try {
			if (FolderExist(_modelFolder)) {
				if (!DeleteFolder(_modelFolder)) {
					this->_logger.LogError("OPENVINOCL::~OPENVINOCL", "Failed to delete OpenVINO Models", __FILE__, __LINE__);
				}
			}
		}
		catch (std::exception& e) {
			this->_logger.LogError("OPENVINOCL::~OPENVINOCL()", "Failed to release OPENVINO Models", __FILE__, __LINE__);
		}
	}
	bool OPENVINOCL::Destroy() {
		try {
			if (FolderExist(_modelFolder)) {
				if (!DeleteFolder(_modelFolder)) {
					this->_logger.LogError("OPENVINOCL::Destroy", "Failed to delete OpenVINO Models", __FILE__, __LINE__);
				}
			}
			return true;
		}
		catch (std::exception& e) {
			this->_logger.LogError("OPENVINOCL::Destroy()", "Failed to release OPENVINO Models", __FILE__, __LINE__);
			return false;
		}
	}
	//private
	void OPENVINOCL::InitialModel() {
		try {
			// Step 1: Initialize OpenVINO Runtime Core
			ov::Core core;

			// Step 2: Load Model	 
			// Get available devices
			std::vector<std::string> available_devices = core.get_available_devices();
			bool device_found = false;

			// Search for NPU
			auto it = std::find(available_devices.begin(), available_devices.end(), "NPU");
			if (it != available_devices.end()) {
				core.set_property("NPU", ov::hint::performance_mode(ov::hint::PerformanceMode::LATENCY));
				core.set_property("GPU", ov::hint::performance_mode(ov::hint::PerformanceMode::LATENCY));
				compiled_model_ = core.compile_model(_modelFilePath, "AUTO:NPU,GPU");
				device_found = true;
				//this->_logger.LogDebug("OPENVINOCL::InitialModel", "Using NPU for inference.", __FILE__, __LINE__);
			}

			// If NPU not found, search for GPU
			if (!device_found) {
				it = std::find(available_devices.begin(), available_devices.end(), "GPU");
				if (it != available_devices.end()) {
					core.set_property("GPU", ov::hint::performance_mode(ov::hint::PerformanceMode::LATENCY));
					compiled_model_ = core.compile_model(_modelFilePath, "GPU");
					device_found = true;
					//this->_logger.LogDebug("OPENVINOCL::InitialModel", "Using GPU for inference.", __FILE__, __LINE__);
				}
			}

			// If GPU not found, search for GPU.0
			if (!device_found) {
				it = std::find(available_devices.begin(), available_devices.end(), "GPU.0");
				if (it != available_devices.end()) {
					core.set_property("GPU", ov::hint::performance_mode(ov::hint::PerformanceMode::LATENCY));
					compiled_model_ = core.compile_model(_modelFilePath, "GPU");
					device_found = true;
					//this->_logger.LogDebug("OPENVINOCL::InitialModel", "Using GPU for inference.", __FILE__, __LINE__);
				}
			}
			// If neither NPU nor GPU found, default to CPU
			if (!device_found) {
				core.set_property("CPU", ov::hint::performance_mode(ov::hint::PerformanceMode::LATENCY));
				compiled_model_ = core.compile_model(_modelFilePath, "CPU");
				//this->_logger.LogDebug("OPENVINOCL::InitialModel", "Using CPU for inference.", __FILE__, __LINE__);
			}

			// Step 3: Create Inference Request
			inference_request_ = compiled_model_.create_infer_request();

		}
		catch (const std::exception& e) {
			this->_logger.LogFatal("OPENVINOCL::InitialModel", e.what(), __FILE__, __LINE__);
		}
	}
}


//std::vector<Object> OPENVINOCL::RunInference(const cv::Mat& input, const std::string& camera_id) {
//	std::lock_guard<std::recursive_mutex> lock(_mutex);
//	std::vector<Object> outputs;

//	if (!_licenseValid) {
//		this->_logger.LogError("OPENVINOCL::RunInference", "Invalid License", __FILE__, __LINE__);
//		return outputs;
//	}
//	if (!_isInitialized) {
//		this->_logger.LogError("OPENVINOCL::RunInference", "Model is not initialized", __FILE__, __LINE__);
//		return outputs;
//	}

//	try {
//		// Step 0: Prepare input
//		if (input.empty()) {
//			this->_logger.LogError("OPENVINOCL::RunInference", "Input image is empty", __FILE__, __LINE__);
//			return outputs;
//		}
//		if ((input.cols < 20) || (input.rows < 20)) return outputs;
//		cv::Mat img = input.clone();
//		cv::resize(img, img, cv::Size(224, 224));
//		cv::Mat letterbox_img = PreProcessing(img);
//		cv::Mat blob = cv::dnn::blobFromImage(letterbox_img, 1.0 / 255.0, cv::Size(224, 224), cv::Scalar(), true);

//		// Step 1: Feed blob to the network
//		auto input_port = compiled_model_.input();
//		ov::Tensor input_tensor(input_port.get_element_type(), input_port.get_shape(), blob.ptr(0));
//		inference_request_.set_input_tensor(input_tensor);

//		// Step 3: Start inference
//		inference_request_.start_async();
//		inference_request_.wait();

//		// Step 4: Get output
//		auto output = inference_request_.get_output_tensor(0);
//		auto output_shape = output.get_shape();

//		if (output_shape.size() < 2) {
//			this->_logger.LogError("OPENVINOCL::RunInference", "Invalid output shape", __FILE__, __LINE__);
//			return outputs;
//		}

//		// Step 5: Post-processing
//		const float* output_buffer = output.data<const float>();
//		std::vector<float> result(output_buffer, output_buffer + output_shape[1]);

//		auto max_idx = std::max_element(result.begin(), result.end());
//		int class_id = static_cast<int>(std::distance(result.begin(), max_idx));
//		float score = *max_idx;
//		int classNameSize = static_cast<int>(_classes.size());
//		Object clsResult;
//		clsResult.classId = class_id;
//		//clsResult.className = (class_id < _classes.size()) ? _classes[class_id] : "";
//		if (!_classes.empty()) {
//			if (clsResult.classId < classNameSize) {
//				clsResult.className = _classes[clsResult.classId];
//			}
//			else {
//				clsResult.className = _classes[classNameSize - 1]; // Use last valid class name if out of range
//			}
//		}
//		else {
//			clsResult.className = "Unknown"; // Fallback if _classes is empty
//		}

//		clsResult.confidence = score;
//		if (input.cols > 20 && input.rows > 20) {
//			clsResult.box = cv::Rect(10, 10, input.cols - 20, input.rows - 20);
//		}
//		else {
//			clsResult.box = cv::Rect(0, 0, input.cols, input.rows);
//		}
//		clsResult.polygon = ANSUtilityHelper::RectToNormalizedPolygon(clsResult.box, input.cols, input.rows);
//		clsResult.cameraId = camera_id;
//		outputs.push_back(clsResult);

//		return outputs;
//	}
//	catch (const std::exception& e) {
//		this->_logger.LogFatal("OPENVINOCL::RunInference", e.what(), __FILE__, __LINE__);
//	}
//	catch (...) {
//		this->_logger.LogFatal("OPENVINOCL::RunInference", "Unknown exception occurred", __FILE__, __LINE__);
//	}

//	return outputs;
//}