Advanced installation guide for Ubuntu - compilation from source files#

The easiest way to install Intel® Deep Learning Streamer (Intel® DL Streamer) Pipeline Framework is installing it from pre-built Debian packages with one-click script. If you would like to follow this way, please go to Install Guide Ubuntu.

The instruction below focuses on installation steps with building Intel® DL Streamer Pipeline Framework from the source code provided in repository.

Step 1: Install prerequisites#

Please go through prerequisites 1 & 2 described in Install Guide Ubuntu

Step 2: Clone Intel® DL Streamer repository#

First, clone this repository into folder ~/intel/dlstreamer_gst:

mkdir -p ~/intel
git clone --recursive https://github.com/dlstreamer/dlstreamer.git ~/intel/dlstreamer_gst
cd ~/intel/dlstreamer_gst

Step 3: Install Intel® Distribution of OpenVINO™ Toolkit#

Follow OpenVINO™ Toolkit instruction guide here to install OpenVINO™ on Linux.

  • Environment: Runtime

  • Operating System: Linux

  • Version: Latest

  • Distribution: OpenVINO™ Archives

After successful OpenVINO™ Toolkit package installation, run the following commands to install OpenVINO™ Toolkit dependencies and enable OpenVINO™ Toolkit development environment:

sudo -E /opt/intel/openvino_2024/install_dependencies/install_openvino_dependencies.sh
source /opt/intel/openvino_2024/setupvars.sh

Install Open Model Zoo tools:

python3 -m pip install --upgrade pip
python3 -m pip install openvino-dev[onnx,tensorflow,pytorch]

Note

Make sure your environment variable $PATH includes ‘$HOME/.local/bin’ .

Step 4: Install Intel® DL Streamer Pipeline Framework dependencies#

Install build dependencies:

# Install dependencies
sudo apt-get update && sudo apt-get install curl wget gpg software-properties-common cmake build-essential libpython3-dev python-gi-dev libopencv-dev jq libgflags-dev libdrm-dev

Download pre-built Debian packages for GStreamer from GitHub Release page <https://github.com/dlstreamer/dlstreamer/releases>. You can manually download all packages from the release page or try to use following command. Install GStreamer from downloaded packages:

  1. For Ubuntu 24.04:

    wget $(wget -q -O - https://api.github.com/repos/dlstreamer/dlstreamer/releases/latest | \
  jq -r '.assets[] | select(.name | contains ("ubuntu_24.04_amd64.deb")) | .browser_download_url')
sudo apt install -y ./intel-dlstreamer-gst*
sudo apt install -y ./intel-dlstreamer-ffmpeg*

  2. For Ubuntu 22.04:

    wget $(wget -q -O - https://api.github.com/repos/dlstreamer/dlstreamer/releases/latest | \
  jq -r '.assets[] | select(.name | contains ("ubuntu_22.04_amd64.deb")) | .browser_download_url')
sudo apt install -y ./intel-dlstreamer-gst*
sudo apt install -y ./intel-dlstreamer-ffmpeg*

Step 5: Install OpenCV#

Download Ninja build system and use it to build OpenCV library:

mkdir -p ~/intel/dlstreamer_gst
cd ~/intel/dlstreamer_gst
sudo apt-get install ninja-build unzip
wget -q --no-check-certificate -O opencv.zip https://github.com/opencv/opencv/archive/4.10.0.zip
unzip opencv.zip && rm opencv.zip && mv opencv-4.10.0 opencv && mkdir -p opencv/build
cd ./opencv/build
cmake -DBUILD_TESTS=OFF -DBUILD_PERF_TESTS=OFF -DBUILD_EXAMPLES=OFF -DBUILD_opencv_apps=OFF -GNinja .. \
&& ninja -j "$(nproc)" && sudo ninja install

Step 6: Install Python dependencies#

If you intend to use Python elements or samples, you need to install the necessary dependencies using the following commands:

  • If you want to set up a local environment follow this step (skip to install all the dependencies globally):

    • Install the virtualenv tool to create isolated Python environments

      # you can install it within the global python interpreter
python3 -m pip install virtualenv
# or you can install it as a user package
python3 -m pip install --user virtualenv

    • Create a virtual environment and activate it

      cd ~/intel/dlstreamer_gst
virtualenv -p /usr/bin/python3 .env3 --system-site-packages
source .env3/bin/activate

  • Install Python requirements:

    cd ~/intel/dlstreamer_gst
python3 -m pip install --upgrade pip
python3 -m pip install -r requirements.txt

Step 7: Install OpenCL™#

Note

Optional step

Run the following commands to install OpenCL™ driver:

# Install
sudo apt-get update && sudo apt-get install -y intel-opencl-icd ocl-icd-opencl-dev opencl-clhpp-headers

Step 8: Install message brokers#

Note

Optional step

Installation of message brokers will enable capability of gvametapublish element to publish inference results to Kafka or MQTT. Note that gvametapublish element will be able to publish inference results to file even if you skip this step.

Run following script to install the message brokers:

cd ~/intel/dlstreamer_gst/scripts/
sudo -E ./install_metapublish_dependencies.sh

In order to enable one of these message brokers in Pipeline Framework, corresponding key should be passed to cmake in configuration step. To enable MQTT please pass -DENABLE_PAHO_INSTALLATION=ON option to cmake. To enable Kafka please pass -DENABLE_RDKAFKA_INSTALLATION=ON option to cmake.

Step 9: Install Intel® oneAPI DPC++/C++ Compiler#

Note

Optional step

OpenCL™ driver installation is required before continuing.

Intel® oneAPI DPC++ Compiler will enable execution of gvawatermark and gvatrack elements on GPU.

# Register Intel® oneAPI APT repository
curl -fsSL https://apt.repos.intel.com/intel-gpg-keys/GPG-PUB-KEY-INTEL-SW-PRODUCTS.PUB | sudo gpg --dearmor --output /usr/share/keyrings/intel-sw-products.gpg
echo "deb [signed-by=/usr/share/keyrings/intel-sw-products.gpg] https://apt.repos.intel.com/oneapi all main" | sudo tee /etc/apt/sources.list.d/intel-oneapi.list

# Install Intel® oneAPI DPC++/C++ Compiler
sudo apt-get update && sudo apt-get install -y intel-oneapi-compiler-dpcpp-cpp intel-level-zero-gpu level-zero-dev

# Set up environment
source /opt/intel/oneapi/compiler/latest/env/vars.sh

Step 10: Install VA-API dependencies#

Note

Optional step

OpenCL™ driver installation is required before continuing.

To enable VA-API preprocessing of Pipeline Framework’s inference elements run the following commands:

sudo apt-get install -y intel-dlstreamer-gst-vaapi libva-dev vainfo intel-media-va-driver-non-free
export LIBVA_DRIVER_NAME=iHD

# When using Media, GPU or NPU devices as non-root user, please configure:
sudo usermod -a -G video <username>
sudo usermod -a -G render <username>

# Check installation
vainfo

The output shouldn’t contain any error messages, iHD driver must be found. If no errors occur, please proceed further to the next step.

If you receive error message like the one below, please reboot your machine and try again.

libva info: va_openDriver() returns -1
vaInitialize failed with error code -1 (unknown libva error), exit

If you receive any other errors, please retry installation process. If it’s not resolved even after re-installation, please submit an issue for support on GitHub here.

Additionally, pass -DENABLE_VAAPI=ON option to cmake in configuration step.

Step 11: Build Intel® DL Streamer Pipeline Framework#

With all dependencies installed, proceed to build Pipeline Framework:

# OpenVINO™ Toolkit environment
source /opt/intel/openvino_2024/setupvars.sh
# GStreamer environment
source /opt/intel/dlstreamer/gstreamer/setupvars.sh
# Intel® oneAPI DPC++/C++ Compiler environment (if installed)
source /opt/intel/oneapi/compiler/latest/env/vars.sh

export CPATH=/opt/intel/dlstreamer/include:${CPATH}

# cmake, remove -DENABLE_VAAPI option if VA-API not installed
mkdir ~/intel/dlstreamer_gst/build
cd ~/intel/dlstreamer_gst/build
cmake -DENABLE_VAAPI=ON ..

# make
make -j

# Setup Intel® DL Streamer Pipeline Framework environment
source ~/intel/dlstreamer_gst/scripts/setup_env.sh

When using Media, GPU or NPU devices as non-root user, please configure:

sudo usermod -a -G video <username>
sudo usermod -a -G render <username>

Intel® DL Streamer Pipeline Framework is now ready to use!

Note

The environment variables are removed when you close the shell. Before each run of Intel® DL Streamer you need to setup the environment with the following 2 scripts as above in this step: source /opt/intel/openvino_2024/setupvars.sh and source ~/intel/dlstreamer_gst/scripts/setup_env.sh. As an option, you can set the environment variables in file ~/.bashrc for automatic enabling.

Note

The manual installation on the host machine may have environment issues. If this happened, please consider other ways. Also, you can create an issue on GitHub here.

Step 12: Intel® DL Streamer installation verification#

Intel® DL Streamer has been installed. Please run the gst-inspect-1.0 gvadetect to confirm that GStreamer and Intel® DL Streamer are running

user@my-host:~/intel/dlstreamer_gst/build$ gst-inspect-1.0 gvadetect

If your can see the documentation of gvadetect element, the installation process is completed.

dev_guide/advanced_install/gvadetect_sample_help.png

Step 13: Next steps - running sample Intel® DL Streamer pipelines#

You are ready to use Intel® DL Streamer. For further instructions to run sample pipeline(s), please go to: Tutorial

* Other names and brands may be claimed as the property of others.