Khadas Docs

Amazing Khadas, always amazes you!

User Tools

Site Tools

Home / products / sbc / edge-2l / npu / rknn-sdk-quick-start
products:sbc:edge-2l:npu:rknn-sdk-quick-start

Differences

This shows you the differences between two versions of the page.

Link to this comparison view

Next revision 		Previous revision
products:sbc:edge-2l:npu:rknn-sdk-quick-start [2025/10/21 23:21]
william created 		products:sbc:edge-2l:npu:rknn-sdk-quick-start [2026/05/31 23:55] (current)
nick [Run RKNN Python demo]
Line 1: Line 1:
-====== RKNN SDK Quick Start ======+====== RKNN SDK quick start ======
  
 ===== Introduction ===== ===== Introduction =====
-This document provides a detailed introduction for beginners on how to quickly use RKNN-Toolkit2 on a computer to perform model conversion and deploy it to a edg2l board using RKNPU2. The examples used in this document are integrated into the RKNN Model Zoo.+This document provides a detailed introduction for beginners on how to quickly use RKNN-Toolkit2 on a computer to perform model conversion and deploy it to a Edge 2L board using RKNPU2. The examples used in this document are integrated into the [[https://github.com/airockchip/rknn_model_zoo | RKNN Model Zoo]].
  
-===== Prepare Development Environment ===== +===== Prepare development environment ===== 
-The version information we are using is as follows: \\ +The version information we are using is as follows:  
-python: 3.8 \\ +  * Host PC Python: 3.8 
-rknn-toolkit2: 2.3.2 \\ +  * rknn-toolkit2: 2.3.2 
-rknn model zoo: 2.3.2 \\ +  * rknn model zoo: 2.3.2 
-==== Download RKNN Related Repositories ==== + 
-It is recommended to create a new directory to store the RKNN repositories. For example, create a folder named "Projects" and place the RKNN-Toolkit2 and RKNN Model Zoo repositories in that directory. Refer to the following commands+==== Download RKNN related repositories - Host PC ==== 
 +It is recommended to create a new directory to store the RKNN repositories. For example, create a folder named ''Projects'' and place the RKNN-Toolkit2 and RKNN Model Zoo repositories in that directory. Refer to the following commands
  
 ```shell ```shell
Line 25: Line 26:
 ``` ```
  
-==== Install the RKNN-Toolkit2 Environment on the Computer ====+==== Install the RKNN-Toolkit2 environment on your Host PC ====
  
-=== Install Python ===+=== Install Python - Host PC ===
 If the Python 3.8 environment is not installed on your system, or if there are multiple versions of Python installed, it is recommended to use Miniforge Conda to create a new Python 3.8 environment. If the Python 3.8 environment is not installed on your system, or if there are multiple versions of Python installed, it is recommended to use Miniforge Conda to create a new Python 3.8 environment.
  
-=== Install Miniforge Conda ===+=== Install Miniforge Conda - Host PC ===
  
 Download the Miniforge Conda installer from the following link: Download the Miniforge Conda installer from the following link:
Line 36: Line 37:
 $ wget -c https://github.com/conda-forge/miniforge/releases/latest/download/Miniforge3-Linux-x86_64.sh $ wget -c https://github.com/conda-forge/miniforge/releases/latest/download/Miniforge3-Linux-x86_64.sh
 ``` ```
 +
 Then, install Miniforge Conda using the following command: Then, install Miniforge Conda using the following command:
 ```shell ```shell
Line 41: Line 43:
 $ bash Miniforge3-Linux-x86_64.sh $ bash Miniforge3-Linux-x86_64.sh
 ``` ```
-=== Create Python Environment Using Miniforge Conda ===+=== Create Python environment using miniforge conda - Host PC === 
 +In the terminal window on the computer, execute the following command to switch to the Miniforge conda base 
 +environment:
 ```shell ```shell
 $ source ~/miniforge3/bin/activate $ source ~/miniforge3/bin/activate
 ``` ```
 +Create a Python 3.8 environment named ''py38Toolkit2.3.2'' using the following command:
 ```shell ```shell
 $ conda create -n py38Toolkit2.3.2 python=3.8 $ conda create -n py38Toolkit2.3.2 python=3.8
 ``` ```
 +Activate the ''py38Toolkit2.3.2'' environment. Subsequently, RKNN-Toolkit2 will be installed in this environment
 ```shell ```shell
 $ conda activate py38Toolkit2.3.2 $ conda activate py38Toolkit2.3.2
 ``` ```
 +=== Install RKNN-Toolkit2 - Host PC ===
 +Install via local wheel packag.
  
 +```shell
 +$ cd rknn-toolkit2/rknn-toolkit2/packages/x86_64/
 +$ pip3 install -r requirements_cp38-2.3.2.txt
 +$ pip3 install ./rknn_toolkit2-2.3.2-cp38-cp38-manylinux_2_17_x86_64.manylinux2014_x86_64.whl
 +```
  
-==== Install Compilation Tools on the Computer ====+=== Check if the RKNN-Toolkit2 environment - Host PC === 
 +Switch to Python interactive mode 
 +```shell 
 +$ python 
 +``` 
 +Import the RKNN class 
 +```shell 
 +$ from rknn.api import RKNN 
 +``` 
 + 
 +==== Install compilation tools on your host PC ==== 
 +=== Android - Install NDK for Android system === 
 +Note: This section is applicable to development boards with the Android system. If the board is running on the 
 +Linux system, please skip this section.\\ 
 +Download the Android NDK from the following link: 
 +```shell 
 +$ wget -c https://dl.google.com/android/repository/android-ndk-r19c-linux-x86_64.zip 
 +``` 
 + 
 +Extract Android NDK files 
 +```shell 
 +$ unzip android-ndk-r19c-linux-x86_64.zip 
 +``` 
 + 
 +==== Linux - Install GCC Cross-Compiler for Linux system ==== 
 +Note: This section is applicable to development boards with the Linux system. If the board is running on the 
 +Android system, please skip this section.\\ 
 +Download the GCC Cross-Compiler from the following link: 
 +```shell 
 +$ wget -c https://releases.linaro.org/components/toolchain/binaries/6.3-2017.05/aarch64-linux-gnu/gcc-linaro-6.3.1-2017.05-x86_64_aarch64-linux-gnu.tar.xz 
 +``` 
 +Extract GCC Cross-Compiler files 
 +```shell 
 +$ tar -xJvf gcc-linaro-6.3.1-2017.05-x86_64_aarch64-linux-gnu.tar.xz 
 +``` 
 + 
 +===== Run example programs ===== 
 + 
 +==== Prepare model ==== 
 +Switch to the ''rknn_model_zoo/examples/yolov5/model'' directory, and execute the ''download_model.sh'' script. This 
 +script will download an available YOLOv5 ONNX model and store it in the current model directory. Refer to the 
 +following commands: 
 + 
 +```shell 
 +$ cd Projects/rknn_model_zoo/examples/yolov5/model 
 +$ chmod a+x download_model.sh && ./download_model.sh 
 +``` 
 + 
 +==== Model conversion ==== 
 +Switch to the ''rknn_model_zoo/examples/yolov5/python'' directory and run the ''convert.py'' script. This script converts 
 +the original ONNX model to the RKNN model. Refer to the following commands: 
 + 
 +```shell 
 +$ cd Projects/rknn_model_zoo/examples/yolov5/python 
 +$ chmod a+x convert.py && python convert.py ../model/yolov5s_relu.onnx rk3576 i8 ../model/yolov5s_relu.rknn 
 +``` 
 + 
 +==== Run RKNN Python demo ==== 
 +=== Android === 
 +Switch to the ''rknn_model_zoo/examples/yolov5/python'' directory, execute the ''yolov5.py'' script, and you can run the 
 +YOLOv5 model on the development board using on-board debugging. Refer to the following command: 
 + 
 +```shell 
 +$ cd Projects/rknn_model_zoo/examples/yolov5/python 
 +$ chmod a+x yolov5.py && python yolov5.py --model_path ../model/yolov5s_relu.rknn --target rk3576 --img_show 
 +``` 
 + 
 +=== Linux === 
 + 
 +Running on a Ubuntu 24.04 system which ship the python version of 3.12. 
 + 
 +Download the model zoo to your Edge 2L: 
 + 
 +```shell 
 +git clone -b v2.3.2 https://github.com/airockchip/rknn_model_zoo.git --depth 1 
 +``` 
 + 
 +Install the python virtual environment: 
 + 
 +```shell 
 +sudo apt update 
 +sudo apt install python3.12-venv -y 
 +``` 
 + 
 +Create the python virtual environment: 
 +```shell 
 +python3 -m venv myenv312 
 +``` 
 + 
 +Activate the python virtual environment: 
 + 
 +```shell 
 +source myenv312/bin/activate 
 +(myenv312) khadas@Khadas:~$ python --version 
 +Python 3.12.3 
 +``` 
 + 
 +Install the python package from ''rknn-toolkit2/rknn-toolkit-lite2/packages/rknn_toolkit_lite2-2.3.2-cp312-cp312-manylinux_2_17_aarch64.manylinux2014_aarch64.whl'' 
 + 
 +```shell 
 +(myenv312) khadas@Khadas:~$ pip install rknn_toolkit_lite2-2.3.2-cp312-cp312-manylinux_2_17_aarch64.manylinux2014_aarch64.whl  
 +``` 
 + 
 +Install the runtime libraries from ''rknn-toolkit2/rknpu2/runtime/Linux/librknn_api/aarch64/librknnrt.so'' 
 + 
 +```shell 
 +sudo cp librknnrt.so /usr/lib/ 
 +``` 
 + 
 +Install the OpenCV: 
 + 
 +```shell 
 +(myenv312) khadas@Khadas:~$ pip install opencv-python 
 +``` 
 + 
 +Copy the converted model in [[#model-conversion | Model conversion]] to your Edge 2L: 
 + 
 +```shell 
 +adb push examples/yolov5/model/yolov5s_relu.rknn /home/khadas/rknn_model_zoo/examples/yolov5/model 
 +``` 
 + 
 +Modify the demo code: 
 + 
 +```diff 
 +diff --git a/py_utils/rknn_executor.py b/py_utils/rknn_executor.py 
 +index 21b4132..46b24d4 100644 
 +--- a/py_utils/rknn_executor.py 
 ++++ b/py_utils/rknn_executor.py 
 +@@ -1,5 +1,5 @@ 
 +-from rknn.api import RKNN 
 +- 
 ++from rknnlite.api import RKNNLite as RKNN 
 ++import numpy as np 
 +  
 + class RKNN_model_container(): 
 +     def __init__(self, model_path, target=None, device_id=None) -> None: 
 +@@ -9,10 +9,7 @@ class RKNN_model_container(): 
 +         rknn.load_rknn(model_path) 
 +  
 +         print('--> Init runtime environment') 
 +-        if target==None: 
 +-            ret = rknn.init_runtime() 
 +-        else: 
 +-            ret = rknn.init_runtime(target=target, device_id=device_id) 
 ++        ret = rknn.init_runtime() 
 +         if ret != 0: 
 +             print('Init runtime environment failed') 
 +             exit(ret) 
 +@@ -33,10 +30,12 @@ class RKNN_model_container(): 
 +         else: 
 +             inputs = [inputs] 
 +  
 ++        inputs = [np.expand_dims(item, axis=0) if len(item.shape) == 3 else item for item in inputs] 
 ++ 
 +         result = self.rknn.inference(inputs=inputs) 
 +      
 +         return result 
 +  
 +     def release(self): 
 +         self.rknn.release() 
 +-        self.rknn = None 
 +\ No newline at end of file 
 ++        self.rknn = None 
 +(myenv312) khadas@Khadas:~/rknn_model_zoo$  
 +``` 
 + 
 +Run the Yolov5 demo: 
 + 
 +```shell 
 +(myenv312) khadas@Khadas:~/rknn_model_zoo/examples/yolov5/python$ export QT_QPA_PLATFORM=xcb 
 +(myenv312) khadas@Khadas:~/rknn_model_zoo/examples/yolov5/python$ python yolov5.py --model_path ../model/yolov5s_relu.rknn --target rk3576 --img_show 
 +``` 
 + 
 +{{:products:sbc:edge-2l:npu:yolov5-linux-demo.webp?600|}} 
 + 
 +Deactivate the python virtual enviranment: 
 + 
 +```shell 
 +(myenv312) khadas@Khadas:~/rknn_model_zoo$ deactivate  
 +``` 
 + 
 + 
 +==== Run RKNN C demo ==== 
 +To run a RKNN C demo, you need to first compile the C/C++ source code into an executable file. After that, push 
 +the executable file, model files, input images, and other related files to the development board. Finally, execute 
 +the executable file on the development board. 
 + 
 +=== Android === 
 +In the ''build-android.sh'' script, add the following command: 
 + 
 +```shell 
 +$ ANDROID_NDK_PATH=Projects/android-ndk-r19c 
 +``` 
 +Then, in the ''rknn_model_zoo'' directory, execute the ''build-android.sh'' script, referring to the following command: 
 + 
 +```shell 
 +$ cd Projects/rknn_model_zoo 
 +$ ./build-android.sh -t rk3576 -a arm64-v8a -d yolov5 
 +``` 
 + 
 +Push Files to the Edge 2L board: 
 +```shell 
 +$ cd Projects/rknn_model_zoo 
 +$ adb root 
 +$ adb push install/rk3576_android_arm64-v8a/rknn_yolov5_demo/ /data/ 
 +``` 
 + 
 +Run the Demo on Edge 2L board: 
 +```shell 
 +$ adb shell 
 +$ cd /data/rknn_yolov5_demo/ 
 +$ export LD_LIBRARY_PATH=./lib 
 +$ ./rknn_yolov5_demo model/yolov5s_relu.rknn model/bus.jpg 
 +``` 
 +=== Linux === 
 +In the ''build-linux.sh'' script, add the following command: 
 + 
 +```shell 
 +$ GCC_COMPILER=Projects/gcc-linaro-6.3.1-2017.05-x86_64_aarch64-linux-gnu/bin/aarch64-linux-gnu 
 +``` 
 +Then, in the ''rknn_model_zoo'' directory, execute the ''build-linux.sh'' script, referring to the following command: 
 + 
 +```shell 
 +$ cd Projects/rknn_model_zoo 
 +$ ./build-linux.sh -t rk3576 -a aarch64 -d yolov5 
 +``` 
 + 
 +Push Files to the Edge 2L board: 
 +```shell 
 +$ cd Projects/rknn_model_zoo 
 +$ adb push install/rk3576_linux_aarch64/rknn_yolov5_demo/ /home/khadas/rknn_yolov5_demo 
 +``` 
 + 
 +Run the Demo on Edge 2L board: 
 +```shell 
 +$ adb shell 
 +$ cd /home/khadas/rknn_yolov5_demo 
 +$ export LD_LIBRARY_PATH=./lib 
 +$ ./rknn_yolov5_demo model/yolov5s_relu.rknn model/bus.jpg 
 +``` 
 + 
 +==== View results ==== 
 +By default, the output image is saved at the path ''rknn_yolov5_demo/out.png''. You can use the adb tool to pull it 
 +from the board to the local machine. In the local computer terminal, execute the following command: 
 +```shell 
 +$ adb pull /home/khadas/rknn_yolov5_demo/out.png . 
 +``` 
 +{{:products:sbc:edge-2l:npu:yolov5-bus-result.webp?600|}}
  
  
-==== Install RKNPU2 Environment on the Board ====+===== See also =====
  
-===== Run Example Programs =====+[[https://github.com/airockchip/rknn-toolkit2/tree/master/doc | RKNN Toolkit2 Documentations]]
  
-===== Important Note ===== 
Last modified: 2025/10/21 23:21 by william

Page Tools

Khadas Site

Khadas Community

Khadas GitHub

Khadas Downloads