Deployment & Inspection

  • Connect to your edge Device (for example using ssh)
  • In the platform, navigate to the Edge Devices overview.
  • Use the corresponding button in the upper right corner to add a new Edge device.
  • Name the device so it can be uniquely identified.
  • After clicking “Add”, you will receive a Docker command. Copy this command and execute it on the Edge device you previously connected to.
  • Verify that the corresponding container is running on the Edge device (for example, using docker container ps).
  • Your Edge device should now appear in the list and be marked as Online.
  • By clicking the Details button, you can view more information about your Edge device and see an overview of the containers already deployed on it. Note that the detail page may take a little longer to load, as the data must be retrieved once directly from the respective Edge device.

  • Switch to the Container Templates view.
  • Create a new template for a Triton Inference Server by clicking the plus icon in the upper right corner.
  • Choose a descriptive template name that allows you to easily identify it later.
  • Use the following Git repository as the source for the container template:
  • 👉 https://github.com/hsos-ai-lab/agri-gaia-inference-container-template.git
  • Specify that the “triton” branch should be used.
  • Select the button to build an inference container based on the container template you created.
  • Assign a descriptive name (lowercase and without special characters) and a tag (e.g., “1.0”).
  • Choose the model you trained and the Edge device you added.
  • The device architecture should be filled in automatically.
  • Navigate to the Container Registry overview.
  • Start the deployment by clicking the “Deploy” button in the row of your predefined container image.
  • Select your Edge device and assign a unique name for the deployment.
  • Add the following port bindings: 8000-8000, 8001-8001, and 8002-8002, then start the deployment by clicking the “Deploy” button.
  • Verify that the deployment has started by opening the running tasks view.
  • Once the deployment is complete, you should find the running transferred container on your Edge device.