add samples

wangrong

2025-03-19 291afb9c81aae1d15d9e3edcea618d0a2e1fc5a6

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
 
#pragma once 
 
#include "../vp_primary_infer_node.h" 
#include "../../objects/vp_frame_target.h" 
 
 
 
namespace vp_nodes { 
    // image segmentation based on Mask RCNN 
    // https://github.com/matterport/Mask_RCNN 
    class vp_mask_rcnn_detector_node: public vp_primary_infer_node 
    { 
    private: 
        /* data */ 
        // names of output layers in mask_rcnn 
        const std::vector<std::string> out_names = {"detection_out_final", "detection_masks"}; 
        float score_threshold = 0.5; 
    protected: 
        // override infer and preprocess as mask_rcnn has a different logic 
        virtual void infer(const cv::Mat& blob_to_infer, std::vector<cv::Mat>& raw_outputs) override; 
        virtual void preprocess(const std::vector<cv::Mat>& mats_to_infer, cv::Mat& blob_to_infer) override; 
 
        virtual void postprocess(const std::vector<cv::Mat>& raw_outputs, const std::vector<std::shared_ptr<vp_objects::vp_frame_meta>>& frame_meta_with_batch) override; 
    public: 
        vp_mask_rcnn_detector_node(std::string node_name,  
                            std::string model_path,  
                            std::string model_config_path = "",  
                            std::string labels_path = "",  
                            int input_width = 416,  
                            int input_height = 416,  
                            int batch_size = 1, 
                            int class_id_offset = 0, 
                            float score_threshold = 0.5, 
                            float scale = 1 / 255.0, 
                            cv::Scalar mean = cv::Scalar(0), 
                            cv::Scalar std = cv::Scalar(1), 
                            bool swap_rb = true); 
        ~vp_mask_rcnn_detector_node(); 
    };     
}