AI Navigator Quick Start Guide Appendix

• A. AI Navigator Uninstallation
• B. How to use Bring Your Own Model Project
• C. Settings required when using other IP addresses of target board
• D. Build the Linux and Use your Custom PCB Board or 3rd Party Board

A. AI Navigator Uninstallation

Follow the steps below to uninstall the plugins, AI Navigator, AI Transfer Learning Tool Plugin, and AI Model Conversion Tool Plugin.

1. Launch e² studio.
2. Click [Help] > [About e² studio] and "About e² studio" will appear.
3. Click [Installation Details].
4. Click the [Installed Software] tab and select "Renesas AI Framework"/"AI Transfer Learning Tool Plugin"/"AI Model Conversion Tool Plugin".
   Note You can select only one plugin at this time. Repeat this procedure for other plugins if you have finished uninstalling one.
5. Click [Uninstall...].
6. Click [Finish] when the uninstallation is complete.
7. The "Software Updates" appears prompting you to restart. Click [Restart Now] to restart e² studio.

Note The Docker containers and the directory that AI Transfer Learning Tool Plugin and AI Model Conversion Tool Plugin installed remains when uninstalling.
If you never use these Docker containers and directories, run the following commands to remove them.

• For AI Transfer Learning Tool Plugin

  #Delete docker container and image AI Transfer Learning Tool Plugin installed.
  docker stop tltdoc
  docker rm tltdoc
  docker rmi tlt_backend:latest
  #Delete the installed AI Transfer Learning Tool Plugin directory as shown below.
  #e.g. if e2 studio is installed in ~/.local/share/renesas/e2_studio
  sudo rm -rf ~/.local/share/renesas/rzv_ai_tlt

• For AI Model Conversion Tool Plugin

  #Delete docker image AI Model Conversion Tool Plugin installed.
  docker rmi rzv2l_ai_sdk_image:latest
  #Delete the installed RZ/V AI TVM Plugin directory as shown below.
  #e.g. if e2 studio is installed in ~/.local/share/renesas/e2_studio
  rm -rf ~/.local/share/renesas/v2l/drp-ai_tvm

B. How to use Bring Your Own Model Project

Bring Your Own Model (hereafter, BYOM) project is an empty project to develop an AI application with own AI models on e² studio.
Please follow the steps if you develop your AI applications by using BYOM project.

Note Bring Your Own Model project does not support RZ/V AI TLT.

1. Import an Bring Your Own Model project

Importing this project is the same as importing other AI application projects. For details, see Step 3: Import an AI Application Project in AI Navigator Quick Start Guide for the details.
In order to import an BYOM project, click [Bring Your Own Model] in the category selection view > [Bring Your Own Model] in the AI application selection view > [Import] in the AI application import view. Be sure to set the directory path of the downloaded and extracted zip file in the Project and AI information view.

2. Convert AI model

This method is the same as for other AI applications. See Option2: Convert AI Model in AI Navigator Quick Start Guide for the details.

3. Edit AI application

To build your AI application source code, you should first import your code into your project as shown below.

• (1) Open Project Explorer, right-click on "src" directory, and click "Import".
• (2) Select "File System".
• (3) Click "Browse" and select the soure directory you want to import.
• (4) Select files you want to import.
• (5) Click [Finish].

Also, change an execution file name to an appropriate one as shown below.

• (1) Right click on the target project in Project Explorer and open "Properties.
• (2) Click [C/C++ Build] -> [Settings] -> [Build Artifact].
• (3) Enter a file name in “Artifact name:”.
• (4) click [Apply and Close].

After that, you can build as in Option3: Edit Application in AI Navigator Quick Start Guide.

4. Run on the board

Before running on the board, you have to change some configuration to run your AI application.

• (1) Edit the copy AI model/application script file
  You need to edit debugCopyFiles.sh in (the path of your project)/ai-navi, which is the script for copying your AI models and applications to the target board.
  Open debugCopyFiles.sh and edit the following definition in lines 50-55 .
  
  • TVM_HOME: The path to your application directory.
  • MODEL_PATH[]: The path to your model directory on e² studio.
  • TARGET_MODEL_PATH: The path to your model directory on the target board.
  • EXE_FILE: Your application file on the target board.



• (2) Change debug configurations
  Select [Debug] > [Debug Configurations...] > [Target Communication Framework] > target project and click [Application] tab. Change the file name specified in “Local File Path:” and “Remote File Path:”.

After that, you can run as in Step 4: Run on the Board in AI Navigator Quick Start Guide.

C. Settings required when using other IP addresses of target board

In AI Navigator, the IP addresses of the target board are set to 192.168.1.11 respectively by default.
If you are using other IP address, you will need to reconfigure some settings as shown below.

• (1) Change the target board IP address setting in the copy process of AI models and AI applications.
  Open /ai-navi/debugCopyFiles.sh in your project directory and set the appropriate target board IP address.
  
  

  IP_ADDRESS="192.168.1.11"

• (2) Reconfigure the debug setting.
  Connect your Linux host PC and the target board.
  Then, click [Run] > [Debug Configurations...] > [Target Communication Framework] > target project and select the [Target] tab.
  In Available targets, select "TFC Agent" whose Host IP address is the target board, and click [Apply] and [Close].
  At this point, check that the Target ID value includes the IP addresses of the target boards.

D. Build the Linux and Use your Custom PCB Board or 3rd Party Board

The current AI Navigator supports the board and the usage described in the Quick Start Guide.
If you want to build the Linux (AI SDK) or develop with your custom PCB board or 3rd party board, please refer to the following pages.

• How to build AI SDK Source Code
  The build step of AI SDK environment.
• RZ/G and RZ/V BSP Porting
  The information in order to port RZ/V Linux environment (includes kernel, u-boot, and so on) into your custom board.