RZ/V2L-EVKIT

RZ/V2L AI Applications is a collection of applications running on the Renesas RZ/V2L vision AI chip. It is available on Renesas' GitHub pages. This tutorial video is based on RZ/V2L AI SDK version 2.10.

Learn more: AI Applications and AI SDK on RZ/V series

Transcript

This video is a tutorial on AI applications.
AI is becoming part of our lives. It has been used in various areas and it will keep spreading in more.
However, it is not easy to implement AI in applications.
To overcome such challenges, Renesas has released AI Applications and AI SDK for RZ/V series.
With these solutions, customers can develop AI Applications for their business easily and quickly.
This video is a tutorial on AI Applications and AI SDK.
It consists of three chapters and we will go through them in this order.

First, we will prepare the necessary environment.
Let's begin with the hardware preparation.
Please obtain the RZ/V2L Evaluation Board Kit. We will explain how to get it.
To get the board, visit Renesas RZ/V AI Web and click here. Click here.
The distributors selling the RZ/V2L Evaluation Board Kit and the remaining stock will be displayed.
Select the distributor and purchase the board.
Once you get the board, please prepare these items. This will complete your hardware preparation.

Next, let's switch to the Ubuntu PC preparation.
We will install the docker engine and Git on the Ubuntu PC.
First, install the docker engine.
From the Renesas RZ/V AI webpages, move to the official Docker page like this.
Type "Ubuntu" in the search window, please select here.
Install the docker engine as instructed here.
Once you downloaded the docker engine, install git in your ubuntu PC.
Open the terminal window. Run the installation commands in the terminal.
You need to set up your username and email address.

You have now completed the preparation of necessary equipment and software.
We can now proceed to "Set up AI SDK".

Next, we will set up AI SDK.
AI SDK is software for running AI Applications quickly and easily on the RZ/V2L Evaluation Board Kit.

We will first obtain AI SDK. To obtain AI SDK, visit the Renesas RZ/V AI webpages and click here.
Next, on this page, click here for the latest version.
Once you've downloaded the AI SDK, let's set it up.

Next, install AI SDK.
The commands can be accessed from the Renesas RZ/V AI webpages like this.
Please see here on getting started.
AI SDK is installed on top of the Docker engine as shown in the picture.
Let's install AI SDK.
Create a working directory. Register the working directory path. Go to the working directory.
Extract AI SDK here. Check the contents of the working directory.
If all directories and files are generated as shown in the log, the extraction was successful.

Then we will set up AI SDK.
Go to the working directory. Install AI SDK by building docker image. Build is completed.
Create new directory for docker container. Run the docker container.
Copy the DRP-AI TVM runtime for later use on board.
AI SDK is installed in docker container, which allows you to build the application.
You can exit the docker container by typing exit command.
AI SDK setup is now completed. The next step is to run AI Application.

Next, to check that the AI SDK has been set up properly, run the AI Application by following these steps.
AI Application is a quick and easy solution to run AI for your own use case.
It uses DRP-AI TVM to accelerate AI processing.
AI Applications can be accessed like this. Please select the category of your interest.
There are many AI Applications.
In addition to these applications, another application is available to check your setup.
It is this application. In this tutorial, we will use it.

We will try building the AI application.
The commands used in this section can be accessed like this.
The commands are described here. Copy and paste to use them.

Let's start.
Go to the working directory. Get the application source code from GitHub. The download is completed.
Then, start the build environment. Register the environment variable. Go to the source code directory.
Create a directory for the build and move to it, and build the source code.
The application build is complete. Check the results of the application build.
If this file is created, it means AI application has been built successfully.
You have completed building the AI Application.
In the next step, the docker container is not used, so please exit the container.

Next, you need to deploy AI application to the board.
Before starting to set up the microSD card, please note that some procedures are required
only when you start using AI SDK or switch to a new version of AI SDK.

First, we will need to create the partitions on the microSD card.
The commands used in this part can be accessed like this.
The commands are described here. Copy and paste to use them.
Regarding the microSD card, please prepare one with a least 4 gigabytes of free space.
Please note that the process explained here will erase all contents stored on your microSD card.

On Linux PC, a microSD card is controlled by the device file name.
In this tutorial, we use "sdb".
Device file name is assigned by the Ubuntu PC system when it recognizes the microSD card.
On your system, the device file name "sdb" may be assigned to other media.
If the "sdb" is assigned to other media, writing to "sdb" will overwrite the data and may destroy it.
In order to avoid system destruction of the media, you must check the device file name of your microSD card.

Now, let's start writing the files to the microSD card.
First, we will check the device file name of the microSD card.
Make sure that you have not inserted the microSD card to the Ubuntu PC and run this command.
Insert the microSD card to the PC and run the same command again.
Compare the results to check the device file name.
Here, the microSD card has this device file name.

Once you know your device file name, check whether current partitions are automatically mounted or not.
If it is already mounted, unmount it since it may cause error when formatting the microSD card.
Here, two partitions are automatically mounted.
Unmount the 1st partition. Unmount the 2nd partition.

Run the fdisk command to create two partitions.
Create a new DOS disklabel. Create a new partition. Select the primary partition.
Specify the 1st partition. Enter the 1st sector. Enter the last sector. Remove the signature.
Create a new partition. Select the primary partition. Specify the 2nd partition.
Enter the 1st sector. Enter the last sector. Remove the signature.
Display partition information. Write the partition information and finish fdisk command.
Reflect the partition updates. Display the microSD partition information.
Format the first partition with ext4. Format the second partition with ext4.

Now you have created the partitions on the microSD card.
Before moving to the next step, you need to eject and insert the microSD card again to mount the newly created partitions.
Run eject command. Remove the microSD card from the PC. Insert the microSD card again.

Next, we will write the Linux files.
The commands used in this part can be accessed like this. The commands are described here. Copy and paste to use them.
Go to the working directory.
To obtain the files, extract this zip file. Confirm the extraction result.
Please check that these files are shown.

Check that you have two partitions on your microSD card.
If the result is shown in the log, you have two partitions. Create a directory for the microSD card.

Mount the partition 1. Copy the Linux Kernel Image to partition 1. Copy the Linux Device Tree File to partition 1.
Copy the Linux kernel files to partition 1. Sync the microSD card to write all data stored in the cache.
Unmount the partition 1.

Mount the partition 2. Next, extract the Linux filesystem to partition 2.
Also, copy the necessary runtime file for AI Application.
Sync the microSD card to write all data stored in the cache.
Unmount the partition 2. Now you have completed writing the Linux files.

Next, we will write the bootloaders.
The commands used in this part can be accessed like this. The commands are described here. Copy and paste to use them.
Go to the bootloader directory.
Check the contents in the bootloader directory. Check that these files are shown.
Write the 1st bootloader to microSD card.
Write the 2nd bootloader to microSD card.
Write the 3rd bootloader to microSD card.
Sync the microSD card to write all data stored in the cache.
Now you have completed writing bootloaders.

Next, we will write the application to microSD card.
The application directory structure will be like this.
The commands used in this part can be accessed like this. The commands are described here. Copy and paste to use them.

We will write the application files to partition 2.
Mount the partition 2. Create the working directory on microSD card.
Start the container. Register the environment variable. Go to the yolo v3 onnx directory.
Download the file from the GitHub for Model Object. Rename the file. Exit the container. Exit the container.
Copy the label file. Then, copy the Model Object directory. Copy the Application binary.
Finally, check that all files are located appropriately.
Then, sync the microSD card to write all data stored in the cache.
Unmount the partition 2. Eject the microSD card.
Now, you have completed the microSD card setup. Remove the microSD card.

Now let's run the AI application. First, we will connect the board and all other equipment.
Insert the microSD card to the board. Change the switch configuration.
Connect the USB hub with the mouse and the keyboard.
Connect the Google Coral camera to the board. The blue part of the cable should be on the top.
Connect the board to the monitor using the micro HDMI cable.
And finally connect the USB Type-C cable to the power port. Two LEDs light up.
Now check that the overall connection looks like this.
Now, we will boot the board.
Press and hold the red power switch for 1 second.
When the third LED lights up, the board will start up.

If you see the log and the yocto screen on your monitor, the board has been booted successfully.
You can now run the AI Application. Let's check the monitor screen.

Click the icon at the top-left corner to open the terminal.
When typing to the terminal, please note that the keyboard is recognized as an English keyboard.
Go to the working directory. Change the permission of the application executable file.
Run the application.

AI Application is started.
The application detects items captured by the Google Coral camera.
You will see bounding boxes and respectively each detected item's details.
To exit the application, press the super key and the tab key simultaneously to go back to the terminal, then press the enter key.
To shutdown the board, enter the shutdown command.
Verify that the power down message is displayed like this.
After the power down log, press and hold the red power switch for 2 seconds.
When the LED turned off, disconnect the USB Type-C cable from the board. Then, disconnect all other cables.
Now you have build and run the AI Application.

Other than the example we have shown here, Renesas has released many other AI applications.
AI applications are grouped by category. Please select the category of your interest. 
You can find them via this webpage.
Please try the one you are interested in.

Please submit your question or request to Renesas.
You can send your questions and requests on AI Applications and AI SDK via Issues on GitHub.

This is the end of the tutorial. Thank you for watching.
For more information, please visit Renesas GitHub Pages.

RZ/V2L AI Applications is a collection of applications running on the Renesas RZ/V2L vision AI chip. It is available on Renesas' GitHub pages.

Learn more: AI Applications and AI SDK on RZ/V series

Transcript

This video is a tutorial for RZ/V2L AI Applications and AI SDK (Software Development Kit). RZ/V2L AI Applications is a collection of applications running on Renesas vision AI chip RZ/V2L. It is available on Renesas' GitHub pages.

Customers will receive the following benefits by using the AI applications. 

• You can evaluate an AI application and use it in your product for free. 
• From various AI applications that Renesas has prepared, you can select an application that matches your business purposes. 
• AI applications have been adjusted to fit the usage situation. It can be used as is without any modification.

AI applications can be quickly and easily run on the RZ/V2L evaluation board by using the AI SDK, which can be obtained free of charge from the Renesas website. The AI SDK includes the software shown in this figure. AI Applications use a USB/MIPI camera with Video for Linux 2 (V4L2). For more information about the software, please refer to the RZ/V2L AI SDK Release Note.

The goals of this tutorial are building an AI Development Environment with RZ/V2L AI SDK and running AI applications prepared on Renesas GitHub Pages. You can achieve these goals by following the get started on Renesas GitHub Pages.

From here, an overview of this tutorial is explained. 
Step 1 describes how to obtain the RZ V2L Evaluation Board kit. 
Step 2 describes the necessary equipment and software. 
In step 3, AI SDK is obtained from the Renesas website and saved on the Ubuntu PC. 
In step 4, AI SDK is extracted. 
In step 5, a Docker image is generated from the Dockerfile included in the AI SDK. The Docker container is built from the docker image. 
In step 6, the application source files obtained from GitHub are built in the container. 
In the first half of step 7, the bootloader files included in the AI SDK are copied from the Ubuntu PC to the Windows PC, and those are written to the RZ/V2L evaluation board using the Windows PC. 
In the second half of step 7, two partitions are created on the microSD card. One is used for the Linux kernel image and the Linux device tree file to boot Linux on the RZ/V2L evaluation board. The other is used for the Linux root file system. 
In step 8, the built application file and the AI model-related files are copied to the Linux root file system. 
In step 9, environment variables are set to the RZ/V2L evaluation board using a Windows PC. Linux system starts up on the RZ/V2L evaluation board using the prepared microSD card. 
Step 10 shows how to run AI Applications on the RZ/V2L evaluation board.

For a better understanding of this tutorial, please refer to the documentation as you follow the instructions in this video. This tutorial is based on version 1.00 of AI SDK. Please refer to the documentation shown above when the version changes.

This step describes how to obtain an evaluation board. 
Click on the "Get RZ/V2L EVK" button.
Click on the cart symbol. The distributors selling the RZ/V2L Evaluation Board Kit and the remaining stock will be displayed. 
Select the distributor and purchase the RZ/V2L Evaluation Board Kit.

This step describes how to obtain the necessary environment. This picture shows all the necessary equipment to run the AI applications.

MIPI Camera Module, micro USB cable, and RZ/V2L Evaluation Board are included in the RZ/V2L Evaluation Board Kit. After obtaining the RZ/V2L Evaluation Board Kit, please prepare the following items.

USB Hub, USB Keyboard, USB Mouse and USB Camera. The USB Camera is optional and is only required for specific applications.

USB Type-C Cable, AC Adaptor and micro HDMI Cable. Linux PC, MicroSD Card and SD Card Reader. Please refer to the documentation (https://renesas-rz.github.io/rzv_ai_sdk/1.00/getting_started.html) for the required Ubuntu version. Linux PC must have internet access.

HDMI Monitor and Windows PC. Windows PC also requires internet access.

This is the list of all necessary equipment explained. Please prepare them.

To run AI applications, the following software must be installed on your Ubuntu PC. The Docker Engine is used to build a development environment for AI applications. Git is used to copy AI applications stored on GitHub.

First, install the Docker engine on your Ubuntu PC. 
Click on "Docker" link. Type "ubuntu" in the search window and select "Install Docker Engine on Ubuntu". 
Install Docker Engine following the content of "Set up and install Docker Engine from Docker’s apt repository".

Then, install git on your Ubuntu PC.
Run the "sudo apt-get update" command in the terminal.
Install git.
Set up your username and email address to complete step 2.

This step describes how to obtain RZ/V2L AI SDK. In this step, the AI SDK is downloaded to the Ubuntu PC from the Renesas website. 
In the documentation step 3, click on the "Download Link" button. 
Click on the account symbol and then click "Log In". 
Enter your email address and password to log in. 
Click on the RZ/V2L AI Software Development Kit. 
Click on the Software License Agreement. 
Check the software license agreement. 
Click on the "Accept and download" button. The download will begin. The AI SDK file size is about 10 GB, which will take time to download.

This step describes how to extract RZ/V2L AI SDK package. In this step, the AI SDK zip file is extracted. 
First, create a working directory. 
Then, register the working directory path as an environment variable. 
Move to the working directory. 
Extract the RZ/V2L AI SDK zip file. 
If all directories and files are generated as shown in the log, the AI SDK zip file has been properly extracted.

This step describes how to set up RZ/V2L AI SDK. In this step, a Docker image is generated from the Dockerfile, and then a Docker container is built from the Docker image. 
Move to the ai_sdk_setup directory. 
Create the Docker image named rzv2l_ai_sdk_image. 
Create a new directory named "data" under ai_sdk_setup directory. 
Create the Docker container named rzv2l_ai_sdk_container. 
When the container is created, the prompt will change. 
In the container, copy the tvm runtime (libtvm_runtime.so) to the data directory (/drp-ai_tvm/data). 
Exit the container.

This step describes how to build RZ/V2L AI Application. In this step, the application source file obtained from GitHub is built in the docker container. The build method presented here is only one example. For details on how to build each AI application, please refer to the documentation (https://renesas-rz.github.io/rzv_ai_sdk/1.00/getting_started.html). AI applications are uploaded to GitHub. In this tutorial, R01_object_detection will be the example application, like a "Hello World". 
Move to the data directory in the host Linux environment. 
Clone the specified repository from GitHub. 
Start the container. 
Then bash is opened to run commands in the container and the prompt changes. 
Register the path to the cloned repository as an environment variable. 
Move to the directory where the source code of the Application is stored. Create the directory for building source code. 
Move to "build" directory. Build the application. 
If a file named object_detection is created, the application build has succeeded. Exit the container to complete step 6.

This step describes how to set up the RZ/V2L Evaluation Board Kit. In the first half of this step, the bootloader files are copied from the Ubuntu PC to the Windows PC, and those are written to the RZ/V2L evaluation board via serial communication. To run AI applications, the following software must be installed on your Windows PC. These software are used for the serial communication between Windows PC and RZ/V2L Evaluation Board. The evaluation board is equipped with a USB UART IC made by FTDI, so the driver software made by FTDI is required. In this tutorial, Tera Term is an example of terminal software. 
In the documentation step 7, click on the address of the FTDI chip. 
Click "here" and the Windows driver installer is downloaded. 
Click on the Installation Guides page. 
Click the Windows 10/11 installation guide. 
Follow the installation guide to install the Windows driver. 
To install Tera term, download it from this website.

In preparation for writing Bootloaders to the evaluation board, copy the three files in the bootloader directory (${WORK}/board_setup/bootloader) to your Windows PC. The process of writing bootloaders needs to be done only once. To write bootloaders, Windows PC and RZ/V2L Evaluation Board need to be connected as shown in this figure.

Connect a Windows PC to the evaluation board using the micro USB Cable. Set SW11 as shown in the screen.

Before connecting the USB type-C cable, the resulting connection should be as shown in the figure. 
Supply power to the board by connecting the USB Type-C cable. 
When the two green LEDs light up, press and hold the red power switch for one second. The LED in the red frame lights up. 
Start tera term on your Windows PC. 
Open the New Connection window and select "Serial" Communication and "USB Serial Port". Click the "OK" button. 
Open the Terminal setup window, check the new line conditions, and click the "OK" button. 
Open the Serial port setup and connection window, verify the serial communication settings, and click the "New setting" button. 
Press the blue reset button. 
On Tera Term, "please send!" message appears. 
Open the Send file window, select the Flash Writer file (*.mot), and click the "Open" button. 
Type "XLS2" command. 
Type the value "11E00". 
Then, type the value "00000". 
When "please send!" message appears, open the Send file window, select bl2_bp-smarc-rzv2l_pmic.srec, and click the "Open" button. 
Type "y" to clear the data. 
Type "XLS2" command to write the next file. 
Type the value "00000". 
Then, type the value "1D200". 
When the "please send!" message appears, open the Send file window, select fip-smarc-rzv2l_pmic.srec, and click the "Open" button. 
Type "y" to clear the data. 
Disconnect serial communication and exit teraterm.

After writing bootloaders, the board must be rebooted. Press and hold the red power switch for 2 seconds. When the LED in the red frame turns off, disconnect the USB type-c cable from the evaluation board. Then remove the micro USB Cable.

In the second part of this step, the microSD card is prepared to boot Linux on the RZ/V2L. Two partitions are created on the microSD card. One is used for the Linux kernel image and the Linux device tree file to boot Linux on the RZ/V2L evaluation board. The other is used for the Linux root file system.

Prepare the microSD card with at least 4 GB of free space. The process described in this step will erase all contents stored on your micro SD card. 
Run the "lsblk" command before inserting the microSD card into the Linux PC. 
Check the command result and then, insert the microSD card into the Linux PC. 
Run the "lsblk" command again. 
Compare the two results to see the device name of the microSD card. 
Run the "df" command to see if the partitions on the microSD card are automatically mounted. If any partitions are automatically mounted, unmount them. 
Run the "fdisk" command to create a partition. 
Enter "o" to create a new DOS disklabel. 
Type "n" to create a new partition. 
Type "p" for partition type. 
Press ENTER key for partition number. 
Press ENTER key for first sector. 
Enter "+500M" for last sector. 
Type "Y" to remove the signature. 
Type "n" to create the second partition. 
Type "p" for partition type. 
Press ENTER key for partition number. 
Press ENTER key for first sector. 
Press ENTER key for last sector. 
Type "Y" to remove the signature. 
Type "p" to display partition information. 
Type "t" to change the partition type. 
Type "1" for partition number. 
Type "b" to specify the type. 
Type "w" to write the partition information and finish fdisk command. 
Run "partprobe" command to reflect the partition updates. 
Display the microSD partition information. 
Format the first partition as FAT32. 
Format the second partition as ext4. 
"Run the "df" command to check if the two partitions are created successfully as shown in the console. 
Mount the microSD card partition 1 to write the necessary files. 
Copy the Linux Kernel Image to the microSD card partition 1. 
Copy the Linux Device Tree File to the microSD card partition 1. 
Run "sync" command. 
Unmount the microSD card partition 1. 
Mount the microSD card partition 2 to write the necessary files. 
Extract the Linux root filesystem to partition 2. 
Copy the tvm runtime to the specified location. 
Run "sync" command. 
Unmount the microSD card partition 2.

Step 8 describes how to deploy the application to the board. In this step, the built application file in step 6 and the object files of the pre-trained AI model are copied to the Linux root file system. At the end of this step, the folder structure of the root file system of the micro SD will look like this figure. 
Insert the micro SD card prepared in step 7 into the Linux PC. 
Mount the partition 2. 
Create an application directory named tvm in the root file system. 
Start the container to copy the necessary files for the AI Applications to the execution environment. 
Set the environment variable. 
Move to the yolov3_onnx directory. 
Download the specified file from GitHub. 
Rename it to deploy.so. 
Exit the container. 
Copy the necessary files to the application directory. 
After copying, verify that the required files are in the proper location. 
Run "sync" command. 
Unmount partition 2. 
Eject the microSD card.

This step describes how to boot the RZ/V2L Evaluation Board Kit. In this step, environment variables are set to the RZ/V2L evaluation board using a Windows PC. Linux system starts up on the RZ/V2L evaluation board using the prepared microSD card.

The process of setting environment variables in U-boot command mode needs to be done only once. If you have done this process, you can skip it.

To set boot parameters, the equipment must be connected to the board as shown in this figure. 
Insert the microSD card prepared in step 8 to the evaluation board. 
Set SW11 as shown on the screen. 
Connect the Windows PC to the evaluation board using the micro USB Cable. 
Connect the Google Coral camera to the evaluation board. The blue part of the cable should be on the upper side. 
Connect an HDMI monitor to the evaluation board using the micro HDMI cable. 
Before connecting the USB type-C cable, the resulting connection should be as shown in the figure. 
Supply power to the board by connecting the USB Type-C cable. 
When the two green LEDs light up, press and hold the red power switch for one second. The LED in the red frame lights up. 
Start Tera Term. 
Configure Tera Term as described in step 7 for serial communication between the Windows PC and the board. 
Press the blue reset button. 
Press Enter key before the countdown reaches zero to enter the U-boot command mode. If you missed it, press the reset button again. 
In U-boot command mode, set the environment variables. Please refer to the documentation and copy & paste the commands. 
When the boot process is complete, a login message will be displayed. " 
Input the following information to verify login. 
After confirmation, shutdown is required to reconnect the cables. 
Enter the Shutdown command. 
Verify that the console displays "Power down". 
Press and hold the red power switch for 2 seconds. When the LED in the red frame turns off, disconnect the USB type-c cable from the evaluation board.

Step 10 describes how to run the application. In this step, the AI application stored on the microSD card is run on the RZ/V2L evaluation board. To run the AI application, the board connection must be changed as shown in this figure. 
Disconnect the micro USB Cable from the evaluation board. 
Connect a USB hub. 
Before connecting the USB type-C cable, the resulting connection should be as shown in the figure. 
Power the board again by connecting the USB type-C cable. 
When the two green LEDs light up, press and hold the red power switch for one second. The LED in the red frame lights up. 
On the HDMI monitor, Yocto Linux starts up. From this point, the screen displays the HDMI monitor until the shutdown process starts. 
Click the icon at the top-left corner to open the terminal. Note that the keyboard connected to the RZ/V2L evaluation board is recognized as an English keyboard even if it is a Japanese keyboard. 
Move to tvm directory. 
Change the permission of the object detection application executable file. 
Run the application. 
The object detection application starts. 
The display shows the image captured by the camera and AI results. 
To exit the application, press the super key (Windows key) and the tab key simultaneously to go back to terminal, then press enter key.

This step describes how to shut down the RZ/V2L Evaluation Board Kit. 
Enter the Shutdown command in the terminal. 
Verify that the "Power down" message appears. 
Press and hold the red power switch for 2 seconds. When the LED in the red frame turns off, disconnect the USB type-c cable from the evaluation board. 
Then, disconnect everything connected to the board.

The next time you boot the RZ/V2L Evaluation Board, you do not need to write bootloaders (in step 7) and set environment variables in U-boot command mode (in step 9). Please start from the state shown in this figure.

For more information, please visit Renesas GitHub Pages.

This video demonstrates the Renesas AI Model Deployer, a GUI-based tool integrated with NVIDIA’s TAO Toolkit. Designed to run locally on standard workstations, it simplifies AI model development and deployment for object detection and image classification at the edge.

Related Resources

• Renesas AI Model Deployer for Vision AI