DashCamNet & PeopleNet Multi-Camera on Xavier NX

Deploys NVIDIA pre-trained DashCamNet (vehicle detection) and PeopleNet (person detection) models on Jetson Xavier NX with multi-camera support using DeepStream and the jetson-multicamera-pipelines project. Achieves ~16.5% CPU for 6 camera streams.

Hardware: Jetson Xavier NX Dev Kit or Jetson SUB Mini PC, USB/CSI cameras, HDMI display Software: JetPack 4.5 or 4.6, NGC account + API key

Execution model

Run one phase at a time. After each phase:

• If output contains [STOP] → stop immediately, consult the failure decision tree
• If output ends with [OK] → tell the user "Phase N complete" and proceed

Phase 1 — prerequisites check (~30 s)

sudo apt-cache show nvidia-jetpack
# Confirm JetPack 4.5 or 4.6

ls /dev/video*
# Confirm cameras connected

[OK] when JetPack 4.5/4.6 confirmed and cameras detected.

Phase 2 — install Docker Engine (~5 min)

sudo apt-get purge docker docker-engine docker.io containerd runc
sudo apt-get update
sudo apt-get install -y ca-certificates curl gnupg lsb-release
curl -fsSL https://download.docker.com/linux/ubuntu/gpg | \
  sudo gpg --dearmor -o /usr/share/keyrings/docker-archive-keyring.gpg
echo "deb [arch=$(dpkg --print-architecture) signed-by=/usr/share/keyrings/docker-archive-keyring.gpg] \
  https://download.docker.com/linux/ubuntu $(lsb_release -cs) stable" | \
  sudo tee /etc/apt/sources.list.d/docker.list > /dev/null
sudo apt-get update
sudo apt-get install -y docker-ce docker-ce-cli containerd.io

Configure non-root access:

sudo groupadd docker
sudo usermod -aG docker $USER

Log out and back in, then verify:

docker run hello-world

[OK] when hello-world runs without sudo.

Phase 3 — install NVIDIA Container Toolkit (~2 min)

distribution=$(. /etc/os-release;echo $ID$VERSION_ID) \
  && curl -s -L https://nvidia.github.io/nvidia-docker/gpgkey | sudo apt-key add - \
  && curl -s -L https://nvidia.github.io/nvidia-docker/$distribution/nvidia-docker.list | \
  sudo tee /etc/apt/sources.list.d/nvidia-docker.list
sudo apt-get update
sudo apt-get install -y nvidia-docker2
sudo systemctl restart docker

[OK] when Docker restarts without errors.

Phase 4 — install NGC CLI (~2 min)

wget -O ngccli_arm64.zip https://ngc.nvidia.com/downloads/ngccli_arm64.zip
unzip -o ngccli_arm64.zip
chmod u+x ngc
echo "export PATH=\"\$PATH:$(pwd)\"" >> ~/.bash_profile
source ~/.bash_profile

Generate an NGC API key at https://catalog.ngc.nvidia.com → Setup → Get API Key

ngc config set
# Enter API key when prompted

[OK] when ngc is configured.

Phase 5 — install TAO Toolkit (~3 min)

sudo apt install -y python3 python3-pip
pip3 install virtualenv
virtualenv venv
source venv/bin/activate
pip3 install nvidia-pyindex
pip3 install nvidia-tao

Verify:

tao --help

If tao not found:

export PATH=$PATH:~/.local/bin
tao --help

[OK] when tao --help shows available tasks.

Phase 6 — install DeepStream 5.1 (~3 min)

Edit apt sources to use r32.5:

sudo nano /etc/apt/sources.list.d/nvidia-l4t-apt-source.list
# Change r32.6 to r32.5 for both lines

sudo apt update
sudo -H pip3 install pyds-ext

[OK] when pyds-ext installs.

Phase 7 — install multicamera pipelines (~5 min)

git clone https://github.com/NVIDIA-AI-IOT/jetson-multicamera-pipelines.git
cd jetson-multicamera-pipelines
bash scripts/install_dependencies.sh
sudo -H pip3 install Cython
sudo -H pip3 install .

[OK] when package installs without errors.

Phase 8 — run multi-camera detection (~2 min)

source scripts/env_vars.sh
cd examples
sudo -H python3 example.py

Edit example.py to match your camera indices:

pipeline = CameraPipelineDNN(
    cameras=[0, 1, 2],  # adjust to your camera device indices
    models=[
        PeopleNet.DLA1,
        DashCamNet.DLA0,
    ],
    save_video=True,
    save_video_folder="/home/$USER/logs/videos",
    display=True,
)

[OK] when multi-camera detection is running with overlays visible.

Failure decision tree

Reference files

• references/source.body.md — Original Seeed wiki with screenshots, NGC setup, and license plate detector example