Skip to main content

RKNN Installation


This document aims to demonstrate how to install the RKNN SDK. For more information, please refer to the RKNN Toolkit2 repository doc directory.

Introduction to RKNN

Rockchip RK3566/RK3568 series, RK3588 series, K3562 series, RV1103/RV1106 series chips are equipped with a neural network processor (NPU). Using RKNN, users can quickly deploy AI models to Rockchip chips for NPU hardware-accelerated inference. To use RKNPU, users need to first use the RKNN-Toolkit2 tool on an x86 computer to convert the trained model into the RKNN format, and then use the RKNN C API or Python API for inference on the development board.

Required Tools:

  • RKNN-Toolkit2 is a software development kit for users to perform model conversion, inference, and performance evaluation on PC and Rockchip NPU platforms.
  • RKNN-Toolkit-Lite2 provides a Python programming interface for Rockchip NPU platforms, helping users deploy RKNN models and accelerate AI applications.
  • RKNN Runtime provides C/C++ programming interfaces for Rockchip NPU platforms, helping users deploy RKNN models and accelerate AI applications.
  • RKNPU kernel driver is responsible for interacting with the NPU hardware.

The overall framework is as follows:


Set up the RKNN Environment

Configure RKNN-Toolkit2 Environment on PC

  • Download the RKNN Repository

    It is recommended to create a directory to store the RKNN repository. For example, create a folder named Projects and place the RKNN-Toolkit2 v1.6.0 and RKNN Model Zoo v1.6.0 repositories under this directory. The commands are as follows:

    # Create Projects folder
    mkdir Projects
    cd Projects

    # Download RKNN-Toolkit2 repository
    git clone -b v1.6.0

    # Download RKNN Model Zoo repository
    git clone -b v1.6.0
  • (Optional) Install Anaconda

    If Python 3.8 (recommended version) is not installed in the system, or if there are multiple Python environments installed simultaneously, it is recommended to use Anaconda to create a new Python 3.8 environment.

    • Install Anaconda

      Execute the following command in the computer's terminal window to check if Anaconda is installed. If Anaconda is already installed, this step can be skipped.

      $ conda --version
      conda 23.10.0

      If "conda: command not found" appears, it means Anaconda is not installed. Please refer to the Anaconda official website for installation.

    • Create a conda environment

      conda create -n rknn python=3.8
    • Activate the conda environment

      conda activate rknn
    • Deactivate the environment

      conda deactivate

Install Dependencies and RKNN-Toolkit2 on PC

  • After activating the conda rknn environment, navigate to the rknn-toolkit2 directory and install dependencies libraries based on your Python version by selecting the corresponding requirements_cpXX.txt file. Then install RKNN-Toolkit2 using the wheel package. The commands are as follows:

    # Navigate to the rknn-toolkit2 directory
    cd Projects/rknn-toolkit2/rknn-toolkit2
    # Choose the appropriate requirements file according to your python version
    pip install -r packages/requirements_cp38-1.6.0.txt -i
    # Choose the appropriate wheel package file according to your python version and processor architecture:
    pip install packages/rknn_toolkit2-1.6.0+81f21f4d-cp38-cp38-linux_x86_64.whl
  • Verify if the installation is successful

    Execute the following command. If no errors occur, it means that the RKNN-Toolkit2 environment is successfully installed.

    $ python3
    >>> from rknn.api import RKNN

Install RKNN Toolkit Lite2 and Its Dependencies on the Board


Radxa official image has RKNPU2 and its dependencies installed by default. Only python3-rknnlite2 needs to be installed. If it doesn't work, try to comment out the command.

sudo apt update
sudo apt install python3-rknnlite2
# sudo apt install rknpu2-rk3588 # For SOC RK3588 series
# sudo apt install rknpu2-rk356x python3-rknnlite2 # For SOC RK356X series

If you are using the CLI version, you can visit the RKNN Toolkit Lite2 deb package download page.