// // Created by DefTruth on 2021/8/1. // #ifndef LITE_AI_ORT_CV_FACEBOXES_H #define LITE_AI_ORT_CV_FACEBOXES_H #include "lite/ort/core/ort_core.h" namespace ortcv { // reference: FaceBoxes.PyTorch python implementation. // https://github.com/zisianw/FaceBoxes.PyTorch/blob/master/layers/functions/prior_box.py class LITE_EXPORTS FaceBoxes : public BasicOrtHandler { public: explicit FaceBoxes(const std::string &_onnx_path, unsigned int _num_threads = 1) : BasicOrtHandler(_onnx_path, _num_threads) {}; ~FaceBoxes() override = default; private: // nested classes struct FaceBoxesAnchor { float cx; float cy; float s_kx; float s_ky; }; private: const float mean_vals[3] = {104.f, 117.f, 123.f}; // bgr order const float scale_vals[3] = {1.f, 1.f, 1.f}; const float variance[2] = {0.1f, 0.2f}; std::vector steps = {32, 64, 128}; std::vector> min_sizes = { {32, 64, 128}, {256}, {512} }; enum NMS { HARD = 0, BLEND = 1, OFFSET = 2 }; static constexpr const unsigned int max_nms = 30000; private: Ort::Value transform(const cv::Mat &mat) override; // void generate_anchors(const int target_height, const int target_width, std::vector &anchors); void generate_bboxes(std::vector &bbox_collection, std::vector &output_tensors, float score_threshold, float img_height, float img_width); // rescale & exclude void nms(std::vector &input, std::vector &output, float iou_threshold, unsigned int topk, unsigned int nms_type); public: void detect(const cv::Mat &mat, std::vector &detected_boxes, float score_threshold = 0.7f, float iou_threshold = 0.45f, unsigned int topk = 400, unsigned int nms_type = NMS::HARD); }; } #endif //LITE_AI_ORT_CV_FACEBOXES_H