Compiling OpenCV and Developing Applications on the RK3576 Board

This article, provided by developer "Duan Dijun," introduces the development and testing of a compilation solution for OpenCV, along with the setup of its environment, specifically tailored for MYIR's MYD-LR3576 development board, which utilizes the Rockchip RK3576 SoC.

MYD-LR3576 Development Board

The RK3576 SoC features the following configurations:

• 4x Cortex-A72 cores (up to 2.2GHz)

• 4x Cortex-A53 cores (up to 1.8GHz)

• NPU (AI acceleration unit)
  - An Independent NPU with a typical computing power of 6 TOPS (INT8)
  - Supports model acceleration for mainstream frameworks such as TensorFlow Lite, PyTorch, and Caffe (requires Rockchip's RKNN Toolkit toolchain).

Therefore, we can run OpenCV code on MYIR's RK3576 development board to perform visual tasks and fully utilize its capabilities. To compile OpenCV, one must first update the software packages and install the necessary dependencies.

sudo apt update

sudo apt upgrade -y

sudo apt install -y build-essential cmake git pkg-config libgtk-3-dev

libavcodec-dev libavformat-dev libswscale-dev libv4l-dev

libxvidcore-dev libx264-dev libjpeg-dev libpng-dev libtiff-dev

gfortran openexr libatlas-base-dev python3-dev python3-numpy

libtbb2 libtbb-dev libdc1394-22-dev libopenexr-dev

libgstreamer-plugins-base1.0-dev libgstreamer1.0-dev

1. Compile the latest version of OpenCV at the board

Clone the OpenCV and OpenCV contrib repositories:

Master Repository (if not cloned)

git clone https://github.com/opencv/opencv.git

cd opencv

git checkout <version, for example 4.9.0> # Optional, specify the version

Contrib Repository (if cloned)

cd /path/to/opencv_contrib # Replace with your contrib path

git checkout <the same version as the master repository>

Create and enter the build folder in the OpenCV main directory:

mkdir build && cd build

Using CMake Configuration (Key Steps):

cmake -D CMAKE_BUILD_TYPE=RELEASE

-D CMAKE_INSTALL_PREFIX=/usr/local

-D INSTALL_C_EXAMPLES=ON

-D INSTALL_PYTHON_EXAMPLES=OFF

-D OPENCV_GENERATE_PKGCONFIG=ON

-D OPENCV_EXTRA_MODULES_PATH=/home/myir/Downloads/opencv-4.11.0/opencv_contrib-4.11.0/modules

-D BUILD_EXAMPLES=ON

-D BUILD_opencv_python3=OFF

-D BUILD_opencv_python2=OFF

-D BUILD_NEW_PYTHON_SUPPORT=OFF

-D WITH_GTK=ON

-D WITH_FFMPEG=ON ..

After all preparations are complete, environmental inspection and configuration work will begin.

After no errors are reported, proceed to the next step.

make -j6

It is not advisable to compile using all cores, as this may result in dependency errors due to varying compilation speeds among threads.

After compilation, run "make install" directly to install.

Detection and Installation

Run the built-in routine.

2. Set up the application

First, insert the USB camera and check if it is recognized in the terminal using the command:

Lsusb

If it is recognized normally, you should see the keyword 'camera' in the device list.

Alternatively, install the Cheese app to verify that the camera is functioning properly.

Create a new build folder within the cpp directory by executing the command `mkdir build`. Subsequently, compile the built-in example program using the provided command:

cd build

cmake ..

make-j8

You can first test if the OpenCV environment is functioning properly.

Enter

./example_opencv to open the above command, proving that there is no problem with the system's cv environment.

Performance Test

In the hog test, the average frame rate of DNN acceleration is about 10 fps with 8 CPUs almost at full load, which is slightly lower than the Raspberry Pi 5.

The average time in the tapi test environment is 68 ms.