Skip to main content

Run Your First Demo

This guide walks you through running your first object detection demo on the Silight AI Camera.

Prerequisites​

Before starting, ensure you have:

  • Completed hardware setup
  • Board powered on and logged in
  • Camera module connected (IMX678)
  • Ethernet cable connected to PC

Step 1: Verify Board is Ready​

Connect to your board via serial console:

picocom /dev/ttyUSB0 -b 115200

Login with:

  • Username: root
  • Password: root

Step 2: Configure Network​

The default board IP is 10.0.0.1. To configure the network, edit /etc/network/interfaces:

vi /etc/network/interfaces

Step 3: Configure Your PC​

On your development PC, add an IP in the same subnet:

# Linux
sudo ip addr add 10.0.0.2/24 dev eth0

# Or configure via Network Manager
nmcli connection modify "Wired connection 1" ipv4.addresses 10.0.0.2/24
nmcli connection up "Wired connection 1"

Verify connectivity:

ping 10.0.0.1

Step 4: Start the Demo​

On Your PC: Start Video Receiver​

Open a terminal on your PC and run:

gst-launch-1.0 -v udpsrc port=5000 address=0.0.0.0 \
  ! application/x-rtp,encoding-name=H264 \
  ! queue ! rtph264depay ! queue ! h264parse ! avdec_h264 \
  ! queue ! videoconvert \
  ! fpsdisplaysink video-sink=autovideosink text-overlay=false sync=false

:::tip GStreamer Installation If gst-launch-1.0 is not found, install GStreamer:

# Ubuntu/Debian
sudo apt-get install gstreamer1.0-tools gstreamer1.0-plugins-base \
  gstreamer1.0-plugins-good gstreamer1.0-plugins-bad \
  gstreamer1.0-libav

# macOS
brew install gstreamer gst-plugins-base gst-plugins-good \
  gst-plugins-bad gst-libav

:::

On the Board: Start Detection​

SSH to the board or use the serial console, then run:

/home/root/apps/case_studies/detection/detection_case_study

You should see output like:

Starting detection pipeline...
Camera: /dev/video0
Streaming to: 10.0.0.2:5000

Step 5: View the Results​

After a few seconds, a video window should appear on your PC showing:

  • Live camera feed from the Silight AI Camera
  • Bounding boxes around detected objects (person, car, dog, etc.)
  • FPS counter in the corner

Demo Screenshot

Troubleshooting​

No Video Appears​

Symptoms: Black screen or frozen video player

Solutions:

  1. Check firewall: sudo ufw allow 5000/udp (Ubuntu)
  2. Verify network: ping 10.0.0.1 from PC
  3. Check camera: v4l2-ctl -d /dev/video0 --all on board
  4. Restart GStreamer receiver on PC

Low FPS​

Symptoms: Choppy video or low frame rate

Solutions:

  1. Close other applications on PC
  2. Use wired Ethernet (not WiFi)
  3. Reduce resolution in detection case study
  4. Check CPU usage: top on board

Next Steps​

Congratulations! You've successfully run your first AI demo on the Silight AI Camera.

Now you can:

Quick Reference​

TaskCommand
Setup IMX678/usr/bin/setup_imx678_sdr.sh
Run detection/home/root/apps/case_studies/detection/detection_case_study
Check camerav4l2-ctl --list-devices
View streamgst-launch-1.0 udpsrc port=5000 ...