rv1126b yolo文档

RKNPU2安装

安装环境

以下开发板安装环境

root@industio: apt-get update
root@industio: apt-get install zlib1g zlib1g-dev libsm6 libgl1-mesa-glx libprotobuf-dev gcc cmake gcc-aarch64-linux-gnu g++-aarch64-linux-gnu

编译 rknn_yolov5_demo

注：下述操作在虚拟机上执行。

进入rknn_yolov5_demo目录，编译rknn_yolov5_demo：

lijs@dell-PowerEdge-R430$ cd <sdk>/external/rknpu2/examples/rknn_yolov5_demo
# 下面指定编译器
#sdk 编译器路径<sdk>/prebuilts/gcc/linux-x86/aarch64/gcc-arm-10.3-2021.07-x86_64-aarch64-none-linux-gnu/bin

lijs@dell-PowerEdge-R430:~/work/ljswork/rockchip/rv1126b_linux6.1/linux6.1/external/rknpu2/examples/rknn_yolov5_demo$ export GCC_COMPILER=/mnt/work2/lijs/ljswork/rockchip/rv1126b_linux6.1/linux6.1/prebuilts/gcc/linux-x86/aarch64/gcc-arm-10.3-2021.07-x86_64-aarch64-none-linux-gnu/bin/aarch64-none-linux-gnu
lijs@dell-PowerEdge-R430:~/work/ljswork/rockchip/rv1126b_linux6.1/linux6.1/external/rknpu2/examples/rknn_yolov5_demo$ ./build-linux.sh -t rv1126b -a aarch64 -b Release
./build-linux.sh -t rv1126b -a aarch64 -b Release
/mnt/work2/lijs/ljswork/rockchip/rv1126b_linux6.1/linux6.1/prebuilts/gcc/linux-x86/aarch64/gcc-arm-10.3-2021.07-x86_64-aarch64-none-linux-gnu/bin/aarch64-none-linux-gnu
===================================
TARGET_SOC=RV1126B
TARGET_ARCH=aarch64
BUILD_TYPE=Release
BUILD_DIR=/mnt/work2/lijs/ljswork/rockchip/rv1126b_linux6.1/linux6.1/external/rknpu2/examples/rknn_yolov5_demo/build/build_RV1126B_linux_aarch64_Release
CC=/mnt/work2/lijs/ljswork/rockchip/rv1126b_linux6.1/linux6.1/prebuilts/gcc/linux-x86/aarch64/gcc-arm-10.3-2021.07-x86_64-aarch64-none-linux-gnu/bin/aarch64-none-linux-gnu-gcc
CXX=/mnt/work2/lijs/ljswork/rockchip/rv1126b_linux6.1/linux6.1/prebuilts/gcc/linux-x86/aarch64/gcc-arm-10.3-2021.07-x86_64-aarch64-none-linux-gnu/bin/aarch64-none-linux-gnu-g++
===================================
-- Configuring done
-- Generating done
-- Build files have been written to: /mnt/work2/lijs/ljswork/rockchip/rv1126b_linux6.1/linux6.1/external/rknpu2/examples/rknn_yolov5_demo/build/build_RV1126B_linux_aarch64_Release
[ 10%] Building CXX object CMakeFiles/rknn_yolov5_video_demo.dir/src/main_video.cc.o
[ 20%] Building CXX object CMakeFiles/rknn_yolov5_video_demo.dir/src/postprocess.cc.o
[ 30%] Building CXX object CMakeFiles/rknn_yolov5_video_demo.dir/utils/mpp_decoder.cpp.o
[ 40%] Building CXX object CMakeFiles/rknn_yolov5_demo.dir/src/main.cc.o
[ 50%] Building CXX object CMakeFiles/rknn_yolov5_video_demo.dir/utils/mpp_encoder.cpp.o
[ 60%] Building CXX object CMakeFiles/rknn_yolov5_video_demo.dir/utils/drawing.cpp.o
[ 70%] Building CXX object CMakeFiles/rknn_yolov5_demo.dir/src/preprocess.cc.o
[ 80%] Building CXX object CMakeFiles/rknn_yolov5_demo.dir/src/postprocess.cc.o
[ 90%] Linking CXX executable rknn_yolov5_video_demo
[ 90%] Built target rknn_yolov5_video_demo
[100%] Linking CXX executable rknn_yolov5_demo
[100%] Built target rknn_yolov5_demo
[ 40%] Built target rknn_yolov5_demo
[100%] Built target rknn_yolov5_video_demo
Install the project...
-- Install configuration: "Release"
-- Installing: /mnt/work2/lijs/ljswork/rockchip/rv1126b_linux6.1/linux6.1/external/rknpu2/examples/rknn_yolov5_demo/install/rknn_yolov5_demo_Linux/./rknn_yolov5_demo
-- Up-to-date: /mnt/work2/lijs/ljswork/rockchip/rv1126b_linux6.1/linux6.1/external/rknpu2/examples/rknn_yolov5_demo/install/rknn_yolov5_demo_Linux/lib/librknnrt.so
-- Up-to-date: /mnt/work2/lijs/ljswork/rockchip/rv1126b_linux6.1/linux6.1/external/rknpu2/examples/rknn_yolov5_demo/install/rknn_yolov5_demo_Linux/lib/librga.so
-- Up-to-date: /mnt/work2/lijs/ljswork/rockchip/rv1126b_linux6.1/linux6.1/external/rknpu2/examples/rknn_yolov5_demo/install/rknn_yolov5_demo_Linux/./model/RV1126B
-- Up-to-date: /mnt/work2/lijs/ljswork/rockchip/rv1126b_linux6.1/linux6.1/external/rknpu2/examples/rknn_yolov5_demo/install/rknn_yolov5_demo_Linux/./model/RV1126B/yolov5s-640-640.rknn
-- Up-to-date: /mnt/work2/lijs/ljswork/rockchip/rv1126b_linux6.1/linux6.1/external/rknpu2/examples/rknn_yolov5_demo/install/rknn_yolov5_demo_Linux/./model/bus.jpg
-- Up-to-date: /mnt/work2/lijs/ljswork/rockchip/rv1126b_linux6.1/linux6.1/external/rknpu2/examples/rknn_yolov5_demo/install/rknn_yolov5_demo_Linux/./model/coco_80_labels_list.txt
-- Installing: /mnt/work2/lijs/ljswork/rockchip/rv1126b_linux6.1/linux6.1/external/rknpu2/examples/rknn_yolov5_demo/install/rknn_yolov5_demo_Linux/./rknn_yolov5_video_demo
-- Up-to-date: /mnt/work2/lijs/ljswork/rockchip/rv1126b_linux6.1/linux6.1/external/rknpu2/examples/rknn_yolov5_demo/install/rknn_yolov5_demo_Linux/lib/librockchip_mpp.so
-- Up-to-date: /mnt/work2/lijs/ljswork/rockchip/rv1126b_linux6.1/linux6.1/external/rknpu2/examples/rknn_yolov5_demo/install/rknn_yolov5_demo_Linux/lib/libmk_api.so
lijs@dell-PowerEdge-R430:~/work/ljswork/rockchip/rv1126b_linux6.1/linux6.1/external/rknpu2/examples/rknn_yolov5_demo$ 

编译完成后压缩，拷贝rknn_yolov5_demo_Linux.zip至开发板系统

cd ~/work/ljswork/rockchip/rv1126b_linux6.1/linux6.1/external/rknpu2/examples/rknn_yolov5_demo/install
zip -r -q -o rknn_yolov5_demo_Linux.zip rknn_yolov5_demo_Linux/

测试运行YOLOv5

图片测试demo

测试demo路径

百度网盘/EVB1126B/EVB1126B-Linux/4.软件资料/工具/rknn_yolov5_demo_Linux.zip

model目录下存放有测试用的jpg图片和rknn测试模型。

执行如下命令：

$ mkdir rknn
$ unzip rknn_yolov5_demo_Linux.zip -d rknn
$ cd rknn/rknn_yolov5_demo_Linux/
$ ./rknn_yolov5_demo model/RV1126B/yolov5s-640-640.rknn model/bus.jpg

执行完毕后会生成一张out.jpg图片，如下：

视频测试demo

$ rknn_yolov5_demo_Linux# ./rknn_yolov5_video_demo ./model/RV1126B/yolov5s-640-640.rknn ./1280x720.h264 264

最终生成 out.h264

# 播放测试
$ gst-play-1.0 out.h264  

yolov11 测试 （Debian11）

yolo11 当前只支持debian版本

#安装opencv，编译工具链等
$ sudo apt update
$ sudo apt install libopencv-dev git make gcc g++ libsndfile1-dev

#获取lubancat配套例程
$ git clone [https://gitee.com/LubanCat/lubancat_ai_manual_code](https://gitee.com/LubanCat/lubancat_ai_manual_code)
$ cd lubancat_ai_manual_code/example/yolo11/cpp

编译前，设置交叉编译工具链路径到当前环境：

$ export PATH=/home/ido/EVB1126B/buildroot/output/rockchip_evb1126b-v1a/host/bin:$PATH
$ vi build-linux.sh

修改build-linux.sh脚本，找到下面代码部分，将编译器名称修改为SDK下builroot的交叉编译工具链名称：

$ ./build-linux.sh -t rk3588 -d    # 这里指定rk3588

#编译出demo   yolov11_image_demo和yolov11_videocapture_demo
$ cd lubancat_ai_manual_code/example/yolov11/cpp/install/rk3588_linux

# 将下载好的yolo11n.rknn 放到开发板上

#测试 摄像头+yolo11 
$ yolo11_videocapture_demo  ，yolo11_videocapture_demo_zero_copy
#摄像头节点 22，30
# 22 为 /dev/video22 也可改成 30 
$ yolo11_videocapture_demo /vendor/etc/firmware/yolo11n.rknn 22
# 或者
$　yolo11_videocapture_demo_zero_copy /vendor/etc/firmware/yolo11n.rknn 22