#include <iostream>
#include "structs.h"
#include <chrono>
#include "cuda_runtime_api.h"
#include "CropLP_logging.h"
#include "CropLP_common.hpp"
#include <math.h>
#define DEVICE 0
#define NET s  // s m l x
#define NETSTRUCT(str) createEngine_##str
#define CREATENET(net) NETSTRUCT(net)
#define STR1(x) #x
#define STR2(x) STR1(x)
// #define USE_FP16  // comment out this if want to use FP16
#define CONF_THRESH 0.51
#define BATCH_SIZE 1
// stuff we know about the network and the input/output blobs

static const int INPUT_H_UNET = 224;
static const int INPUT_W_UNET = 224;
static const int OUTPUT_SIZE_UNET = 224*224;

using namespace nvinfer1;
static Logger gLogger;

struct DetectionCropLP{        
    float mask[INPUT_W_UNET * INPUT_H_UNET * 1];
};

class CropLP{
    private:
        std::string model_path = "/home/thai/py_script/MTA/unet-segment/unet/license_ckpt_crop/22-0.9823-0.0604.trt";
        char *trtModelStream{nullptr};
        size_t size{0};
        IRuntime* runtime{nullptr};
        ICudaEngine* engine{nullptr};
        IExecutionContext* context{nullptr};
        const char* INPUT_BLOB_NAME_UNET = "input_1";
        const char* OUTPUT_BLOB_NAME_UNET = "conv2d_23";
        

    public:
        static CropLP &getInst()
        {
            static CropLP instance;
            return instance;
        };
        CropLP();
        void _init();
        float eculidDistance(cv::Point p1, cv::Point p2);
        void validate_point(cv::Point2f &tl, cv::Point2f &tr, cv::Point2f &br, cv::Point2f &bl, cv::Mat img_ori);
        cv::Mat preprocess_img(cv::Mat& img);
        void doInference(IExecutionContext& context, float* input, float* output, int batchSize);
        void process_cls_result(DetectionCropLP &res, float *output);
        float sigmoid(float x);
        cv::Mat segmentation(cv::Mat img);
        cv::Mat Crop(cv::Mat mask, cv::Mat img_ori, std::string mode);
        std::vector<cv::Mat> predict(cv::Mat inputImage, std::string mode);
};