 |
Darknet/YOLO v3.0-149-gb11c9d5
Object Detection Framework
|
|
|
Darknet/YOLO v3.0-149-gb11c9d5
Object Detection Framework
|
|
A place to store other details related to the neural network which we cannot easily add to the usual Darknet::Network structure. More...
#include "darknet_network.hpp"
Public Member Functions | |
| NetworkDetails () | |
| Default constructor. | |
Public Attributes | |
| bool | annotate_draw_bb |
| Whether bounding boxes are drawn when annotating images. | |
| bool | annotate_draw_label |
| Whether text labels are drawn above bounding boxes when annotating images. | |
| float | bounding_boxes_corner_roundness |
The "roundness" of the corners when bounding_boxes_with_rounded_corners is set to true. | |
| bool | bounding_boxes_with_rounded_corners |
| Whether bounding boxes should use rounded corners. | |
| std::vector< cv::Scalar > | class_colours |
| BGR colours to use for each class. | |
| VStr | class_names |
| The name to use for every object class. | |
| cv::HersheyFonts | cv_font_face |
| The OpenCV built-in font to use when generating text, such as the labels above bounding boxes. | |
| double | cv_font_scale |
| The OpenCV font scale to use when generating text, such as the labels above bounding boxes. | |
| int | cv_font_thickness |
| The OpenCV font thickness to use when generating text, such as the labels above bounding boxes. | |
| cv::LineTypes | cv_line_type |
| The OpenCV line type to use when drawing lines such as bounding boxes. | |
| float | detection_threshold |
| Object detection threshold to apply. | |
| bool | fix_out_of_bound_normalized_coordinates |
| Fix out-of-bound values for objects near the edges of images. | |
| float | non_maximal_suppression_threshold |
| Non maximal suppression threshold to apply. | |
| std::vector< cv::Scalar > | text_colours |
| BGR colours to use for the label text. | |
| std::filesystem::path | cfg_path |
| Filename used to load the neural nework. | |
| std::filesystem::path | names_path |
| Filename used to load the neural nework. | |
| std::filesystem::path | weights_path |
| Filename used to load the neural nework. | |
A place to store other details related to the neural network which we cannot easily add to the usual Darknet::Network structure.
These are typically C++ objects, or things added post Darknet V3 (2024-08).
| Darknet::NetworkDetails::NetworkDetails | ( | ) |
Default constructor.
| bool Darknet::NetworkDetails::annotate_draw_bb |
Whether bounding boxes are drawn when annotating images.
Default is true.
| bool Darknet::NetworkDetails::annotate_draw_label |
Whether text labels are drawn above bounding boxes when annotating images.
Default is true.
| float Darknet::NetworkDetails::bounding_boxes_corner_roundness |
The "roundness" of the corners when bounding_boxes_with_rounded_corners is set to true.
Default is 0.5.
| bool Darknet::NetworkDetails::bounding_boxes_with_rounded_corners |
Whether bounding boxes should use rounded corners.
Default is false (meaning square bounding boxes).
| std::filesystem::path Darknet::NetworkDetails::cfg_path |
Filename used to load the neural nework.
| std::vector<cv::Scalar> Darknet::NetworkDetails::class_colours |
BGR colours to use for each class.
| VStr Darknet::NetworkDetails::class_names |
The name to use for every object class.
Will always match the number of classes in the neural network.
| cv::HersheyFonts Darknet::NetworkDetails::cv_font_face |
The OpenCV built-in font to use when generating text, such as the labels above bounding boxes.
Default is cv::HersheyFonts::FONT_HERSHEY_PLAIN.
| double Darknet::NetworkDetails::cv_font_scale |
The OpenCV font scale to use when generating text, such as the labels above bounding boxes.
Default is 1.0.
| int Darknet::NetworkDetails::cv_font_thickness |
The OpenCV font thickness to use when generating text, such as the labels above bounding boxes.
Default is 1.
| cv::LineTypes Darknet::NetworkDetails::cv_line_type |
The OpenCV line type to use when drawing lines such as bounding boxes.
Possible values include cv::LineTypes::LINE_4, cv::LineTypes::LINE_8, and cv::LineTypes::CV_LINE_AA. LINE_4 is the fastest but lowest quality, while LINE_AA (anti-alias) is the slowest with highest quality. Default is cv::LineTypes::LINE_4.
| float Darknet::NetworkDetails::detection_threshold |
Object detection threshold to apply.
Default is 0.25.
| bool Darknet::NetworkDetails::fix_out_of_bound_normalized_coordinates |
Fix out-of-bound values for objects near the edges of images.
Default is true.
| std::filesystem::path Darknet::NetworkDetails::names_path |
Filename used to load the neural nework.
| float Darknet::NetworkDetails::non_maximal_suppression_threshold |
Non maximal suppression threshold to apply.
Default is 0.45.
| std::vector<cv::Scalar> Darknet::NetworkDetails::text_colours |
BGR colours to use for the label text.
| std::filesystem::path Darknet::NetworkDetails::weights_path |
Filename used to load the neural nework.
1.9.8