Skip to content

yolov11 瑞芯微 rknn 板端 C++部署,使用平台 rk3588。

License

Notifications You must be signed in to change notification settings

xiyang18/yolov11_dfl_rknn_Cplusplus

 
 

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

7 Commits
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

yolov11_dfl_rknn_Cplusplus

yolov11 rk3588 部署版本,将DFL放在后处理中,转换工具版本 rknn_toolkit2-2.2.0-cp38-cp38-manylinux_2_17_x86_64.manylinux2014_x86_64.whl.

导出onnx、转rknn流程说明【yolov11 部署瑞芯微rk3588、RKNN部署工程难度小、模型推理速度快】

编译和运行

1)编译

cd examples/rknn_yolov11_demo_dfl_open

bash build-linux_RK3588.sh

2)运行

cd install/rknn_yolo_demo_Linux

./rknn_yolo_demo 

注意:修改模型、测试图像、保存图像的路径,修改文件为src下的main.cc


int main(int argc, char **argv)
{
    char model_path[256] = "/home/zhangqian/rknn/examples/rknn_yolov11_demo_dfl_open/model/RK3588/yolov11n_80class_ZQ.rknn";
    char image_path[256] = "/home/zhangqian/rknn/examples/rknn_yolov11_demo_dfl_open/test.jpg";
    char save_image_path[256] = "/home/zhangqian/rknn/examples/rknn_yolov11_demo_dfl_open/test_result.jpg";

    detect(model_path, image_path, save_image_path);
    return 0;
}

测试效果

onnx 测试效果

image

rk3588上测试效果

images

把板端模型推理和后处理时耗也附上,供参考,使用的芯片rk3588,模型输入640x640。

image

tenorRT cuda 版本参考(cuda 实现后处理中部分最耗时处理)

yolov11_tensorRT_postprocess_cuda

About

yolov11 瑞芯微 rknn 板端 C++部署,使用平台 rk3588。

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages

  • C 89.0%
  • Go 8.8%
  • Shell 2.0%
  • Makefile 0.2%