Skip to main content

Quick Start

Overview​

This chapter provides a quick guide to using the NG4500 series products.

Initial System Startup Configuration​

Refer to the above content, complete the hardware connection, power to NG4500, according to the following prompts to complete the system's first initial configuration.

1. User Configuration​

  1. Check accept license and continue.

  1. Set the language and continue.

  1. Configure the keyboard layout and continue.

  1. Set the timezone (e.g., Asia/Shanghai) and continue.

  1. Configure 'Your name' and 'password', then continue.

System Configuration

  1. Finally, wait for the system installation to complete and restart automatically.

2. System Status & Version Confirmation​

  • Use the ​​Jtop​​ tool to monitor system resources:

    # Install Jtop
    sudo apt update
    sudo apt install python3-pip
    sudo pip3 install -U jetson-stats

    # Restart the service
    sudo systemctl restart jtop.service

    # Run Jtop
    sudo jtop

    This will display real-time resource usage, including CPU, memory, GPU, disk, and fan status.And confirm the current system is installed TensorRT, CUDA version.

Model Deployment​

The following will introduce how to deploy and run the NanoOWL real - time visual inference model on the AIBOX. Taking NG4511 as an example, an imx219 camera is used.

Hardware Requirements​

ComponentRequirement
DeviceJetson Orin（Nano / NX / AGX）
Memory≥ 8GB (Larger models require more memory)
Storage Space≥ 10GB (Depending on the model size)
GPUNVIDIA GPU supporting CUDA

Software Environment​

  • ​​​JetPack Version​​：
    • JetPack 5 (L4T r35.x)
    • JetPack 6 (L4T r36.x) (Default version of the system)

Environment Preparation​

  1. Hardware connection: Connect the IMX219 camera (Note: The metal contacts should face upwards).

  2. Install dependencies

sudo apt update
sudo apt-get install -y docker.io
sudo apt-get install -y nvidia-container-toolkit
sudo apt-get install nvidia-jetpack
  1. Install jetson - containers
# Get the source code
git clone https://github.com/dusty-nv/jetson-containers

# Install dependencies
bash jetson-containers/install.sh
  1. Start the deployment and automatically pull/build the nanoowl container. (Note: After obtaining the container, the container will be started)
cd jetson-containers/
jetson-containers run --workdir /opt/nanoowl $(autotag nanoowl)

This command will automatically detect your hardware configuration and pull or build the appropriate container image.

Running Example​

  1. Check the camera device
ls /dev/video*

  1. Run the nanoowl container and complete the following configurations

    • Start the docker and enter the test case path
    cd jetson-containers/
    jetson-containers run --workdir /opt/nanoowl $(autotag nanoowl)

    # Enter the test case path
    cd /opt/nanoowl/examples/tree_demo
    • Install dependencies
    apt update
    apt-get install vim
    pip install aiohttp
    • Modify the /opt/nanoowl/examples/tree_demo file
    # Comment out the following line of code
    camera = cv2.VideoCapture(CAMERA_DEVICE)

    # Replace it with the following content
    gst_pipeline = (
        "nvarguscamerasrc ! "
        "video/x-raw(memory:NVMM), width=(int)1280, height=(int)720, format=(string)NV12, framerate=(fraction)30/1 ! "
        "nvvidconv ! "
        "video/x-raw, format=(string)BGRx ! "
        "videoconvert ! "
        "video/x-raw, format=(string)BGR ! "
        "appsink"
    )
    camera = cv2.VideoCapture(gst_pipeline, cv2.CAP_GSTREAMER)

  2. Start the terminal and run the test case

    python3 tree_demo.py --camera 0 --resolution 1920x1080 ../../data/owl_image_encoder_patch32.engine

  3. Results

    • Open the browser and enter the IP address of the current NG4511 device, such as http://<ip address>:7860.
    • Enter any content you want to recognize, such as
      • [a face [a nose, an eye, a mouth]]
      • [a table [a keyboard, a pen, a mouse]]