common.hpp

#ifndef RETINAFACE_COMMON_H_
#define RETINAFACE_COMMON_H_
#include <opencv2/opencv.hpp>
#include <dirent.h>
#include "NvInfer.h"
#include "decode.h"

using namespace nvinfer1;

#define CHECK(status) \
    do\
    {\
        auto ret = (status);\
        if (ret != 0)\
        {\
            std::cerr << "Cuda failure: " << ret << std::endl;\
            abort();\
        }\
    } while (0)

static inline cv::Mat preprocess_img(cv::Mat& img, int input_w, int input_h) {
    int w, h, x, y;
    float r_w = input_w / (img.cols*1.0);
    float r_h = input_h / (img.rows*1.0);
    if (r_h > r_w) {
        w = input_w;
        h = r_w * img.rows;
        x = 0;
        y = (input_h - h) / 2;
    } else {
        w = r_h * img.cols;
        h = input_h;
        x = (input_w - w) / 2;
        y = 0;
    }
    cv::Mat re(h, w, CV_8UC3);
    cv::resize(img, re, re.size(), 0, 0, cv::INTER_LINEAR);
    cv::Mat out(input_h, input_w, CV_8UC3, cv::Scalar(128, 128, 128));
    re.copyTo(out(cv::Rect(x, y, re.cols, re.rows)));
    return out;
}

static inline int read_files_in_dir(const char *p_dir_name, std::vector<std::string> &file_names) {
    DIR *p_dir = opendir(p_dir_name);
    if (p_dir == nullptr) {
        return -1;
    }

    struct dirent* p_file = nullptr;
    while ((p_file = readdir(p_dir)) != nullptr) {
        if (strcmp(p_file->d_name, ".") != 0 &&
            strcmp(p_file->d_name, "..") != 0) {
            //std::string cur_file_name(p_dir_name);
            //cur_file_name += "/";
            //cur_file_name += p_file->d_name;
            std::string cur_file_name(p_file->d_name);
            file_names.push_back(cur_file_name);
        }
    }

    closedir(p_dir);
    return 0;
}

static inline cv::Rect get_rect_adapt_landmark(cv::Mat& img, int input_w, int input_h, float bbox[4], float lmk[10]) {
    int l, r, t, b;
    float r_w = input_w / (img.cols * 1.0);
    float r_h = input_h / (img.rows * 1.0);
    if (r_h > r_w) {
        l = bbox[0] / r_w;
        r = bbox[2] / r_w;
        t = (bbox[1] - (input_h - r_w * img.rows) / 2) / r_w;
        b = (bbox[3] - (input_h - r_w * img.rows) / 2) / r_w;
        for (int i = 0; i < 10; i += 2) {
            lmk[i] /= r_w;
            lmk[i + 1] = (lmk[i + 1] - (input_h - r_w * img.rows) / 2) / r_w;
        }
    } else {
        l = (bbox[0] - (input_w - r_h * img.cols) / 2) / r_h;
        r = (bbox[2] - (input_w - r_h * img.cols) / 2) / r_h;
        t = bbox[1] / r_h;
        b = bbox[3] / r_h;
        for (int i = 0; i < 10; i += 2) {
            lmk[i] = (lmk[i] - (input_w - r_h * img.cols) / 2) / r_h;
            lmk[i + 1] /= r_h;
        }
    }
    if(l<0){
        l=0;
    }
    if(t<0){
        t=0;
    }

    return cv::Rect(l, t, r-l, b-t);
}

static float iou(float lbox[4], float rbox[4]) {
    float interBox[] = {
        std::max(lbox[0], rbox[0]), //left
        std::min(lbox[2], rbox[2]), //right
        std::max(lbox[1], rbox[1]), //top
        std::min(lbox[3], rbox[3]), //bottom
    };

    if(interBox[2] > interBox[3] || interBox[0] > interBox[1])
        return 0.0f;

    float interBoxS = (interBox[1] - interBox[0]) * (interBox[3] - interBox[2]);
    return interBoxS / ((lbox[2] - lbox[0]) * (lbox[3] - lbox[1]) + (rbox[2] - rbox[0]) * (rbox[3] - rbox[1]) -interBoxS + 0.000001f);
}

static bool cmp(const decodeplugin::Detection& a, const decodeplugin::Detection& b) {
    return a.class_confidence > b.class_confidence;
}

static inline void nms(std::vector<decodeplugin::Detection>& res, float *output, float nms_thresh = 0.4) {
    std::vector<decodeplugin::Detection> dets;
    for (int i = 0; i < output[0]; i++) {
        if (output[15 * i + 1 + 4] <= 0.1) continue;
        decodeplugin::Detection det;
        memcpy(&det, &output[15 * i + 1], sizeof(decodeplugin::Detection));
        dets.push_back(det);
    }
    std::sort(dets.begin(), dets.end(), cmp);
    for (size_t m = 0; m < dets.size(); ++m) {
        auto& item = dets[m];
        res.push_back(item);
        //std::cout << item.class_confidence << " bbox " << item.bbox[0] << ", " << item.bbox[1] << ", " << item.bbox[2] << ", " << item.bbox[3] << std::endl;
        for (size_t n = m + 1; n < dets.size(); ++n) {
            if (iou(item.bbox, dets[n].bbox) > nms_thresh) {
                dets.erase(dets.begin()+n);
                --n;
            }
        }
    }
}

// Load weights from files
// TensorRT weight files have a simple space delimited format:
// [type] [size] <data x size in hex>
static inline std::map<std::string, Weights> loadWeights(const std::string file) {
    std::cout << "Loading weights: " << file << std::endl;
    std::map<std::string, Weights> weightMap;

    // Open weights file
    std::ifstream input(file);
    assert(input.is_open() && "Unable to load weight file.");

    // Read number of weight blobs
    int32_t count;
    input >> count;
    assert(count > 0 && "Invalid weight map file.");

    while (count--)
    {
        Weights wt{DataType::kFLOAT, nullptr, 0};
        uint32_t size;

        // Read name and type of blob
        std::string name;
        input >> name >> std::dec >> size;
        wt.type = DataType::kFLOAT;

        // Load blob
        uint32_t* val = reinterpret_cast<uint32_t*>(malloc(sizeof(val) * size));
        for (uint32_t x = 0, y = size; x < y; ++x)
        {
            input >> std::hex >> val[x];
        }
        wt.values = val;
        
        wt.count = size;
        weightMap[name] = wt;
    }

    return weightMap;
}

static inline Weights getWeights(std::map<std::string, Weights>& weightMap, std::string key) {
    if (weightMap.count(key) != 1) {
        std::cerr << key << " not existed in weight map, fatal error!!!" << std::endl;
        exit(-1);
    }
    return weightMap[key];
}

static inline IScaleLayer* addBatchNorm2d(INetworkDefinition *network, std::map<std::string, Weights>& weightMap, ITensor& input, std::string lname, float eps) {
    float *gamma = (float*)weightMap[lname + ".weight"].values;
    float *beta = (float*)weightMap[lname + ".bias"].values;
    float *mean = (float*)weightMap[lname + ".running_mean"].values;
    float *var = (float*)weightMap[lname + ".running_var"].values;
    int len = weightMap[lname + ".running_var"].count;

    float *scval = reinterpret_cast<float*>(malloc(sizeof(float) * len));
    for (int i = 0; i < len; i++) {
        scval[i] = gamma[i] / sqrt(var[i] + eps);
    }
    Weights scale{DataType::kFLOAT, scval, len};
    
    float *shval = reinterpret_cast<float*>(malloc(sizeof(float) * len));
    for (int i = 0; i < len; i++) {
        shval[i] = beta[i] - mean[i] * gamma[i] / sqrt(var[i] + eps);
    }
    Weights shift{DataType::kFLOAT, shval, len};

    float *pval = reinterpret_cast<float*>(malloc(sizeof(float) * len));
    for (int i = 0; i < len; i++) {
        pval[i] = 1.0;
    }
    Weights power{DataType::kFLOAT, pval, len};

    weightMap[lname + ".scale"] = scale;
    weightMap[lname + ".shift"] = shift;
    weightMap[lname + ".power"] = power;
    IScaleLayer* scale_1 = network->addScale(input, ScaleMode::kCHANNEL, shift, scale, power);
    assert(scale_1);
    return scale_1;
}

#endif