yolov9-onnx-opencv

• yolov9-onnx-opencv
  • Requirements
  • Installation
  • Examples
    • Arguments
    • Inference on Image
    • Inference on Video
• References:

yolov9 导出onnx 模型，C++ 使用opencv dnn 预测

Requirements

请自行编译

OpenCV 4.9.0

yaml-cpp

Installation

git clone https://github.com/WeiZhenOoooo/yolov9-onnx-opencv.git
cd yolov9-onnx-opencv

修改CMakeLists.txt，line 6 和 line 7

set(OpenCV_DIR "D:/CredLib_DIR/opencv47/build")
set(Yaml_Cpp_DIR "D:/CredLib_DIR/yaml-cpp")

将OpenCV_DIR和Yaml_Cpp_DIR的路径修改为本机的实际路径，CMake编译

将yaml-cpp.dll复制到可执行文件yolo-onnx-opencv.exe的所在目录，准备执行

Examples

Arguments

E:\code\yolov9-onnx-opencv\cmake-build-release>.\yolo-onnx-opencv.exe -h
yolo onnx opencv dnn pred description
Usage: .\yolo-onnx-opencv.exe [OPTIONS]

Options:
  -h,--help                   Print this help message and exit
  -w,--weights TEXT           onnx model path
  -c,--config TEXT            train yolo yaml path, to find classes definition
  -i,--input TEXT             pred input path
  -z,--size INT               input size(size * size), default:640 * 640
  -t,--threshold FLOAT        threshold score, default: 0.5
  --image,--no-image{false}   Image inference mode
  --video,--no-video{false}   Video inference mode
  --gpu,--no-gpu{false}       Cuda inference mode

如果要使用--gpu参数，请自行编译opencv GPU版本

Inference on Image

.\yolo-onnx-opencv.exe --weights ..\weights\yolov9-c.onnx -c ..\example\metadata.yaml -i ..\example\000000000625.jpg --image

Inference on Video

.\yolo-onnx-opencv.exe --weights ..\weights\yolov9-c.onnx -c ..\example\metadata.yaml -i ..\example\road.mp4 --video

References:

• YOLOv9 : https://github.com/WongKinYiu/yolov9