Skip to main content

DeepStream

This guide explains how to install and run NVIDIA DeepStream SDK on Jetson Orin devices. DeepStream supports GPU-accelerated AI video analytics pipelines and is highly optimized for Jetson's CUDA/NvMedia platform.

1. Overview​

  • Real-time video analytics SDK provided by NVIDIA
  • Accelerated with TensorRT and CUDA
  • Supports multi-stream AI inference and object tracking
  • Input sources include RTSP, USB, CSI cameras, and local video files
  • Built-in object detection, classification, and tracking capabilities

This guide covers:

  • Installation methods (.deb package and Docker)
  • Running sample pipelines
  • Integrating custom models
  • Docker usage (including jetson-containers)
  • Common issues and tips

overview

2. System Requirements​

Hardware​

ComponentMinimum Requirement
DeviceJetson Orin Nano / NX / AGX
RAM≥ 8GB
Storage≥ 10GB

Software​

  • JetPack 6.1 GA or later (L4T ≥ R36.4)
  • Ubuntu 20.04 / 22.04
  • CUDA, TensorRT, cuDNN (included in JetPack)
  • Docker (optional, for containerized deployment)

3. Installing DeepStream​

glib Migration​

To migrate to a newer glib version (e.g., 2.76.6), follow these steps:

Prerequisites: Install the following packages:

sudo pip3 install meson  
sudo pip3 install ninja

Compilation and installation steps:

git clone https://github.com/GNOME/glib.git  
cd glib  
git checkout <glib-version-branch>  
# e.g., 2.76.6  
meson build --prefix=/usr  
ninja -C build/  
cd build/  
sudo ninja install

Verify the installed glib version:

pkg-config --modversion glib-2.0

Dependency Installation​

sudo apt update  
sudo apt install -y \  
  libssl1.1 \  
  libgstreamer1.0-0 \  
  gstreamer1.0-tools \  
  gstreamer1.0-plugins-good \  
  gstreamer1.0-plugins-bad \  
  gstreamer1.0-plugins-ugly \  
  gstreamer1.0-libav \  
  libgstrtspserver-1.0-0 \  
  libjansson4 \  
  libyaml-cpp-dev

Install librdkafka​

  1. Clone the librdkafka repository from GitHub:
git clone https://github.com/confluentinc/librdkafka.git
  1. Configure and build the library:
cd librdkafka  
git checkout tags/v2.2.0  
./configure --enable-ssl  
make  
sudo make install
  1. Copy the generated libraries to the DeepStream directory:
sudo mkdir -p /opt/nvidia/deepstream/deepstream/lib  
sudo cp /usr/local/lib/librdkafka* /opt/nvidia/deepstream/deepstream/lib  
sudo ldconfig

Method 1: Install via SDK Manager​

  1. Download and install SDK Manager from the NVIDIA official website.

  2. Connect the device: Use a USB-C cable to connect the Jetson Orin device to the host computer.

  3. Launch SDK Manager: Run the sdkmanager command on the host and log in with your NVIDIA developer account.

  4. Select target hardware and JetPack version: Choose the corresponding Jetson Orin device and appropriate JetPack version in SDK Manager.

  5. Enable DeepStream SDK: Check the DeepStream SDK option under "Additional SDKs."

  6. Begin installation: Follow the prompts to complete the installation.

Method 2: Using DeepStream Tar Package​

  1. Download DeepStream SDK: Visit the NVIDIA DeepStream download page and download the DeepStream SDK tar package for Jetson (e.g., deepstream_sdk_v7.1.0_jetson.tbz2).

  2. Extract and install:

sudo tar -xvf deepstream_sdk_v7.1.0_jetson.tbz2 -C /  
cd /opt/nvidia/deepstream/deepstream-7.1  
sudo ./install.sh  
sudo ldconfig

Method 3: Using DeepStream Debian Package​

  1. Download DeepStream Debian: Visit the DeepStream Debian download page and download the DeepStream SDK tar package for Jetson (e.g., deepstream-7.1_7.1.0-1_arm64.deb).

  2. Install:

sudo apt-get install ./deepstream-7.1_7.1.0-1_arm64.deb

Method 4: Using Docker​

  1. Install Docker and NVIDIA Container Toolkit: Ensure Docker and NVIDIA Container Toolkit are installed.

  2. Pull the DeepStream Docker image:

docker pull nvcr.io/nvidia/deepstream-l4t:6.1-samples
  1. Run the container:
docker run -it --rm --runtime=nvidia \  
  -v /tmp/.X11-unix:/tmp/.X11-unix \  
  -e DISPLAY=$DISPLAY \  
  nvcr.io/nvidia/deepstream-l4t:6.1-samples

Alternatively, use the community-maintained jetson-containers:

jetson-containers run dusty-nv/deepstream

Installation Verification​

Check version information:

deepstream-app --version-all

Expected output:

deepstream-app version 7.1.0  
DeepStreamSDK 7.1.0  
CUDA Driver Version: 12.6  
CUDA Runtime Version: 12.6  
TensorRT Version: 10.3  
cuDNN Version: 9.0  
libNVWarp360 Version: 2.0.1d3

4. Running Examples​

Step 1: Run Default Example​

  1. Navigate to the configs/deepstream-app directory on the dev kit:
cd /opt/nvidia/deepstream/deepstream-7.1/samples/configs/deepstream-app
  1. Run the reference application:
# deepstream-app -c <path_to_config_file>  
deepstream-app -c source30_1080p_dec_infer-resnet_tiled_display_int8.txt

This command will pop up a video window displaying real-time detection results:

deepstream_app_5x8

Step 2: Using USB or CSI Camera​

Modify the input section in the configuration file:

[source0]  
enable=1  
type=1  
camera-width=1280  
camera-height=720  
camera-fps-n=30

Then run:

deepstream-app -c <your_camera_config>.txt

🎥 USB cameras use type=1, while CSI cameras use nvarguscamerasrc.

Step 3: Using RTSP Stream​

Use the following configuration snippet:

[source0]  
enable=1  
type=4  
uri=rtsp://<your-camera-stream>

Step 4: Video Detection​

Navigate to the example folder:

cd /opt/nvidia/deepstream/deepstream-7.1/sources/apps/sample_apps/deepstream-test1

Compile the source code:

sudo make CUDA_VER=12.6

Run:

./deepstream-test1-app dstest1_config.yml

deepstream_od

For more source examples, see /opt/nvidia/deepstream/deepstream/sources.

5. Integrating Custom Models​

DeepStream supports integrating custom models via TensorRT or ONNX.

Step 1: Model Conversion​

Use trtexec or tao-converter:

trtexec --onnx=model.onnx --saveEngine=model.engine

Step 2: Update Configuration File​

[primary-gie]  
enable=1  
model-engine-file=model.engine  
network-type=0

For more DeepStream TAO examples, visit https://github.com/NVIDIA-AI-IOT/deepstream_tao_apps.

6. More Examples​

deepstream_python_apps
deepstream_python

7. Tips and Troubleshooting​

IssueSolution
No display in DockerMount X11 socket and set DISPLAY variable
Low frame rateUse INT8 engine or reduce input resolution
USB camera not detectedCheck devices with v4l2-ctl --list-devices
GStreamer errorsVerify plugin installation or reflash JetPack
RTSP latency/dropped framesSet drop-frame-interval=0 and latency=200

8. Appendix​

Key Paths​

PurposePath
Sample config files/opt/nvidia/deepstream/deepstream/samples/configs/
Model engine files/opt/nvidia/deepstream/deepstream/models/
Log directory/opt/nvidia/deepstream/logs/
DeepStream CLI tool/usr/bin/deepstream-app

References​