1. Browse for latest engineering release from StarFive at Microsoft OneDrive link: https://debian.starfivetech.com/
2. Navigate towards SD card section and download the Debian image pre-built by StarFive.
3. Download BalenaEtcher application.
4. Open BalenaEtcher software, in the meantime, insert a 32-GB micro-SD card into the laptop using USB micro-SD card reader. Ensure the selected SD storage is the targeted card.
5. Extract the .img file from the downloaded .zip file from Step 2. Upload the image file to BalenaEtcher and start the flash task.
6. After finish writing the disk image, a successfully flash message appears further indicates that the 32-GB micro-SD card is ready to be ejected from laptop and to be inserted into SD card slot of VisionFive 2.

Table 1: Boot Modes Settings

1. Connect Raspberry Pi Official USB-C Power Supply into VisionFive 2 USB C port. For first time setup, it is essential to connect VisionFive2 SBC to monitor through High-Definition Multimedia Interface (HDMI) cable.
2. There are 4 types of boot modes, namely 1-bit QSPI Nor Flash, UART, eMMC and SDIO3.0 based on Table 1. Select the boot mode SDIO with Rapid General-Purpose Input/Output Drivers (RGPIO) configuration as followed: RGPIO_1: 0 (LOW) and RGPIO_0: 1 (HIGH).

3. A Gnome login interface appears which prompts user to enter the credentials as follows. Open Terminal Emulator after successful login. Username: user Password: starfive

1. Ensure USB WiFi adaptor is plugged into the peripheral I/O port of VisionFive2 SBC. Navigate to the network section in setting and connect to a known WiFi. Akin to command 'ipconfig' applied in Windows OS, command 'nmcli -p device show' used to show the IP address of VisionFive 2 SBC. Record the IP address '192.168.137.244'. Note the IP address may vary upon connection to different WiFi network.
2. Enter privileged mode (root).
3. Install UltraVNC viewer on client laptop. Back to VisionFive 2 SBC, install a standalone TigerVNC server.
4. Exit privileged mode and install the MATE Desktop Environment.
5. When setting up a VNC, it is essential for user to set a VNC password. In this context, the VNC password is set as 'starfive'.
6. Initiate the VNC server with MATE desktop, screen resolution [1920x1080], display number 1.
7. Return to client laptop, insert the IP address recorded from STEP 1 into the section, with a format (Server's hostname or IP address):(display number). Press connect and enter password 'starfive' for input authentication. A successful login will be verified by the pop up of Debian OS home page.
8. After every usage, it is optional for user to either manually kill the TigerVNC server or perform shutdown at start icon to close the connection between client laptop with VisionFive2 SBC.

1.To visualize the comprehensive report about various aspects of your system, including hardware configuration, system resources, network information, and more, users may pass command

    $sudo apt-get install inxi
    $inxi --full

2. It is essential to check the usable space on memory compartment (32-GB SD card) and expand the roots partition. If the available usable space of SD card at /dev/mmcblk1p4 is mismatched with the expected specification (usually smaller), several commands are needed to expand the unused space on the SD card. This is important to enable user to install any relevant libraries for future works.

1.At Terminal (non-root mode), install the essential packages provided by StarFive. In this context, user will obtain browsers such as Firefox and Chromium, VLC and FFmpeg as media player, and others packages include node.js, v8, libsdl2-dev, GStreamer, v4l2test, Libreoffice, QT and NW.js. Some of these essential packages provided by StarFive are not available to download via apt/apt-get. To improve overall user experience, install vim and nautilus (Gnome File Viewer) and upgrade several existing packages. Alternatively, install all these libraries simply via commands below:

    $wget https://github.com/starfive-tech/Debian/releases/download/v0.8.0-engineering-release-wayland/install_package_and_dependencies.sh
    $chmod +x install_package_and_dependencies.sh
    $nano install_package_and_dependencies.sh
    $sudo ./install_package_and_dependencies.sh

Table 2: Essential Packages provided by StarFive.

2. Install essential python libraries via command "pip install" or "sudo apt-get install". These libraries support the the operations of OpenCV Legacy Trackers, YOLOv7 and YOLOv8. Alternatively, install all these libraries simply via command.

    $pip install -r requirements.txt
   

Table 3: Essential Python Libraries supporting OpenCV Legacy Trackers, YOLOv7 and YOLOv8.

    $lsusb
    $sudo chmod 666 /dev/video4
    $v4l2-ctl --list-devices

3. To test the camera accessibility, you may activate through ffplay via command

    $ffplay -f v4l2 -framerate 30 -video_size 640x480 /dev/video4
   

or

    Open Python IDLE > Open file "Camera Test.py" > Run

    $nano vnc_connect.sh

2. In the script, insert the code to setup TigerVNC server session from Step 3 and code to setup camera accessibility from Step 5. Save and exit the script environment xia key "^X".

    sleep 10
    tigervncserver -xstartup /usr/bin/mate-session -geometry 1920x1080 -localhost no :1
    
    VIDEO_DEVICE="/dev/video4"
    
    # Check if the video device exists
    if [ -e "$VIDEO_DEVICE" ]; then
        echo "Video device $VIDEO_DEVICE found."
    
        # Perform sudo chmod 666 on the video device
        sudo chmod 666 "$VIDEO_DEVICE"
        echo "Permissions changed successfully."
    else
        echo "Video device $VIDEO_DEVICE not found."
    fi
   

3. Restart the VisionFive2 SBC. Unplug the HDMI cable from the monitor, mouse and keyboard from the USB ports.

4. Open the UltraVNC application on local host computer, insert the IP address along with port number 1, eg. "192.168.137.244:1". A successful connection will be indicated by the pop up window that prompts user to insert password of that server. A MATE Desktop Environment will be displayed upon successful password authentication.

    yolov7 > YOLOv7_Pipe_Tracker.ipynb
    yolov8 > YOLOv8_Pipe_Tracker.ipynb

2. After running all the codes line in the files at Google Colab, navigate to the files tab section at the Google Colab, searching for PyTorch trained model file (.pt) generated at the directories below. Remember to save the directories at Google Drive via function "Mount Drive".

    For YOLOv7, run/train/exp/best.pt
    For YOLOv8, ultralytics/yolo/v8/detect/best.pt
   

1. For CV2 legacy tracker, open Python IDLE > Open file "CV2 Legacy Tracker.py".
2. At line 11, define the camera source, in this case, "/dev/video4".
3. Run the python code via key F5.
4. A window will appear to prompt user to draw a blue bounding box to select the object to track.
5. Visualise the resource usage using glances, glances is a debian resource monitoring tool.

1. For YOLOv7, open Python IDLE > Open file "yolov7/detect.py"
2. At line 185, define the camera source, in this case, "/dev/video4".
3. At line 184, ensure your PyTorch weight file, in this case, r"/home/user/Documents/FYP_19000760/yolov7/runs/train/exp3/weights/best.pt", is saved at the same directory as the "detect.py".
4. Run the python code via key F5.
5. Visualise the resource usage using glances, glances is a debian resource monitoring tool.

1. For YOLOv8, open Python IDLE > Open file "yolov8/YOLOv8 Live Webcam Tracker.py"
2. At line 90, define the camera source, in this case, "/dev/video4".
3. At line 95, ensure your PyTorch weight file, in this case, "bestn.pt", is saved at the same directory as the "YOLOv8 Live Webcam Tracker.py".
4. Run the python code via key F5.
5. Visualise the resource usage using glances, glances is a debian resource monitoring tool.