216 lines
7.0 KiB
C++
216 lines
7.0 KiB
C++
#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;
|
|
}
|
|
}
|
|
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
|
|
|