This guide provides step-by-step instructions for installing VMSS 2.0 Docker for K26 on KV260 or KR260 Kria SOM, using the Kria Ubuntu Desktop 22.04 LTS Image. This tutorial has been tested on both KV260 and KR260 with the image flashed using Balena Etcher on a Windows 11 machine.
- Kria SOM (KV260 or KR260) with Kria Ubuntu Desktop 22.04 LTS Image.
- 32GB storage or more (We used 128GB SanDisk Extreme PLUS 200MB/s Read SD card).
- Follow the initial setup instructions from Xilinx: Setting up the SD Card Image.
NOTE: On preparation of the Kria SOM, ensure that the date is correct otherwise the license may register as expired:
ubuntu@kria:~$ date
Thu Apr 25 03:02:42 UTC 2024
Run the following command in your terminal to download and execute the installation script:
wget -qO aup_vmss_setup.sh https://raw.githubusercontent.com/auperatech/VMSS2.0/main/setup/K260_Kria_SOM/aup_vmss_setup.sh && bash aup_vmss_setup.sh
Follow the on-screen prompts to:
- Specify the full path of the shared directory for Docker container access. Leave blank to use the current directory.
- Enter the name for your Docker container. Leave blank to use the default name
vmss_docker
.
In real scenarios, you would need to reboot, turn off and on the SOM for multiple occasions. rebooting will stop the docker. In this case, it is important to not run the setup script again. Instead just run the following command:
sudo xmutil unloadapp; sudo xmutil loadapp kv260-smartcam
After the installation, you can access your Docker container's interactive bash session with the following command:
sudo docker container exec -ti vmss_docker bash
Replace vmss_docker
with your specified container name if you did not use the default name when prompted for container name in the installation step.
Now that you have completed the installation you can follow our tutorial to run examples and learn about VMSS2.0.
VMSS2.0 utility consists of two major modules, Aupera Video AI Server(AVAS) and Aupera Video AI Client(AVAC). AVAS is the server that runs and manages various AI tasks, and AVAC is a web cloud application that allows users to use a friendly GUI to connect to AVAS. It only takes several simple steps to launch AVAS inside the docker and play with our AVAC after the docker is successfully built. Let's play!
NOTE: The following section assumes that you already entered the docker by the command listed above.
First, please run a single script to start AVAS.
start.sh skip_check
Then, with a single command to get the unique serial number(sn) of your device that we prepared.
cat /opt/aupera/avas/etc/kria_som_sn.txt
Next, create your own account and sign in to our Web Cloud.
Sign up for an account at: https://auperatechnologies.com. Once you sign in with your new account, you will be asked to add a video stream. Please click the Add Stream button to add video streams you would like to use, or you could directly use the demo video streams we prepared for testing as listed below:
rtsp://vmss.auperatechnologies.com:554/crowd
rtsp://vmss.auperatechnologies.com:554/car
rtsp://vmss.auperatechnologies.com:554/retail
rtsp://vmss.auperatechnologies.com:554/imagenet
rtsp://vmss.auperatechnologies.com:554/face
Last, add the device with the unique serial number and start running tasks.
Go to the AI Hosts tab in the navigation bar. Click the Add AI Host button to add a host. Choose the Add by Serial Number tab, enter an AI Host Name of your choice, and enter the unique serial number from the last step in AI Host S/N input.
After device verification and the device is successful, you can go to the AI Tasks section to create your tasks and start to play!
NOTE: For a video and more detailed document to guide you through AVAC and creating tasks, please refer to video guide avac user guide document