Advanced Installation - Compilation From Source#

NOTE: Installation of Deep Learning Streamer Pipeline Framework from pre-built Debian packages using one-click script is the easiest approach.

The instructions below are intended for building Deep Learning Streamer Pipeline Framework from the source code provided in

Open Edge Platform repository.

NOTE: The main branch currently targets a nightly (pre-release) build of OpenVINO™ GenAI, not an official stable release. The exact nightly version/date used below may become outdated over time; check the OPENVINO_GENAI_VERSION build argument in docker/ubuntu/ubuntu24.Dockerfile for the version currently used in CI, and substitute it below if it differs.

Step 1: Install prerequisites (only for Ubuntu)#

Follow the instructions in the prerequisites section.

Step 2: Install build dependencies#

sudo apt-get update && \
sudo apt-get install -y wget vainfo xz-utils python3-pip python3-gi gcc-multilib libglib2.0-dev \
    flex bison autoconf automake libtool libogg-dev make g++ libva-dev yasm libglx-dev libdrm-dev \
    python-gi-dev python3-dev unzip libgflags-dev libcurl4-openssl-dev \
    libgirepository1.0-dev libx265-dev libx264-dev libde265-dev gudev-1.0 libusb-1.0 nasm python3-venv \
    libcairo2-dev libxt-dev libgirepository1.0-dev libgles2-mesa-dev wayland-protocols \
    libssh2-1-dev cmake git valgrind numactl libvpx-dev libopus-dev libsrtp2-dev libxv-dev \
    linux-libc-dev libpmix2t64 libhwloc15 libhwloc-plugins libxcb1-dev libx11-xcb-dev \
    ffmpeg libpaho-mqtt-dev libopencv-dev libpostproc-dev libavfilter-dev libavdevice-dev \
    libswscale-dev libswresample-dev libavutil-dev libavformat-dev libavcodec-dev libtbb12 libxml2-dev \
    ocl-icd-opencl-dev
sudo apt-get update && \
sudo apt-get install -y wget vainfo xz-utils python3-pip python3-gi gcc-multilib libglib2.0-dev \
    flex bison autoconf automake libtool libogg-dev make g++ libva-dev yasm libglx-dev libdrm-dev \
    python-gi-dev python3-dev unzip libgflags-dev \
    libgirepository1.0-dev libx265-dev libx264-dev libde265-dev gudev-1.0 libusb-1.0 nasm python3-venv \
    libcairo2-dev libxt-dev libgirepository1.0-dev libgles2-mesa-dev wayland-protocols libcurl4-openssl-dev \
    libssh2-1-dev cmake git valgrind numactl libvpx-dev libopus-dev libsrtp2-dev libxv-dev \
    linux-libc-dev libpmix2 libhwloc15 libhwloc-plugins libxcb1-dev libx11-xcb-dev \
    ffmpeg libpaho-mqtt-dev libpostproc-dev libavfilter-dev libavdevice-dev \
    libswscale-dev libswresample-dev libavutil-dev libavformat-dev libavcodec-dev libtbb12  libxml2-dev ocl-icd-opencl-dev \
    opencl-headers
sudo dnf install -y \
    https://download1.rpmfusion.org/free/fedora/rpmfusion-free-release-$(rpm -E %fedora).noarch.rpm \
    https://download1.rpmfusion.org/nonfree/fedora/rpmfusion-nonfree-release-$(rpm -E %fedora).noarch.rpm
sudo dnf install -y wget libva-utils xz python3-pip python3-gobject gcc gcc-c++ glibc-devel glib2-devel \
    flex bison autoconf automake libtool libogg-devel make libva-devel yasm mesa-libGL-devel libdrm-devel \
    python3-gobject-devel python3-devel tbb gnupg2 unzip opencv-devel gflags-devel openssl-devel openssl-devel-engine \
    gobject-introspection-devel x265-devel x264-devel libde265-devel libgudev-devel libusb1 libusb1-devel nasm python3-virtualenv \
    cairo-devel cairo-gobject-devel libXt-devel mesa-libGLES-devel wayland-protocols-devel libcurl-devel which \
    libssh2-devel cmake git valgrind numactl libvpx-devel opus-devel libsrtp-devel libXv-devel paho-c-devel \
    kernel-headers pmix pmix-devel hwloc hwloc-libs hwloc-devel libxcb-devel libX11-devel libatomic intel-media-driver
sudo dnf install -y uuid libuuid-devel openssl-devel gcc gcc-c++ make curl ca-certificates librdkafka-devel libva-devel alsa-lib-devel unzip glibc libstdc++ libgcc cmake sudo pkgconf pkgconf-pkg-config ocl-icd-devel libva-intel-media-driver python3-devel libXaw-devel ncurses-devel libva2 intel-compute-runtime intel-opencl intel-level-zero-gpu intel-ocloc-devel nasm

Step 3: Set up a Python environment#

For crating wirtual enviroment uv tool is recomended, and needs to be instaled:

curl -LsSf https://astral.sh/uv/install.sh | sh

Create a Python virtual environment and install required Python packages:

NOTE: Ubuntu 22.04’s default Python (3.10) is not supported by the OpenVINO™ GenAI Python bindings, so install Python 3.11 with uv and use it for the virtual environment below. Ubuntu 24.04 and Ubuntu 26.04 can use their default system Python versions. Build native extensions such as PyGObject inside the same environment so their Python ABI matches the interpreter used by DL Streamer’s Python elements (e.g. gvapython). On Ubuntu 22.04, add the uv-managed Python library directory to the dynamic linker cache so libgstpython.so can load libpython.

UV_PYTHON_PREFERENCE=only-system uv venv --python python3 ~/python3venv
source ~/python3venv/bin/activate

uv pip install meson==1.4.1 ninja==1.11.1.1
uv python install 3.11

python_bin="$(uv python find 3.11)"
python_libdir="$("$python_bin" -c 'import sysconfig; print(sysconfig.get_config_var("LIBDIR"))')"
echo "$python_libdir" | sudo tee /etc/ld.so.conf.d/uv-python.conf >/dev/null
sudo ldconfig

UV_PYTHON_PREFERENCE=managed uv venv --python 3.11 ~/python3venv
source ~/python3venv/bin/activate

uv pip install meson==1.4.1 ninja==1.11.1.1

Step 4: Clone Deep Learning Streamer repository#

cd ~
git clone --branch main --recursive https://github.com/open-edge-platform/dlstreamer.git
cd dlstreamer

Step 5: Install OpenVINO™ GenAI#

On Ubuntu and EMT, the OpenVINO™ GenAI archive bundles the full OpenVINO™ C++ runtime alongside the GenAI libraries, so this step alone is enough to build DL Streamer - no separate “plain” OpenVINO™ Toolkit install is needed on those platforms. Fedora 41 instead installs the official OpenVINO™ Toolkit 2026.3.1 package from Intel’s YUM repository, as in the Fedora 41 Dockerfile. The separate OpenVINO™ GenAI archive is required if you want to use the gvagenai element, which needs the OpenVINO GenAI archive package.

wget -O- https://storage.openvinotoolkit.org/repositories/openvino_genai/packages/pre-release/2026.5.0.0rc2/openvino_genai_ubuntu24_2026.5.0.0rc2_x86_64.tar.gz | tar -xz &&
sudo mv openvino_genai_ubuntu24_2026.5.0.0rc2_x86_64 /opt/intel/openvino_genai
source /opt/intel/openvino_genai/setupvars.sh
wget -O- https://storage.openvinotoolkit.org/repositories/openvino_genai/packages/pre-release/2026.5.0.0rc2/openvino_genai_ubuntu22_2026.5.0.0rc2_x86_64.tar.gz | tar -xz &&
sudo mv openvino_genai_ubuntu22_2026.5.0.0rc2_x86_64 /opt/intel/openvino_genai
source /opt/intel/openvino_genai/setupvars.sh
wget -O- https://storage.openvinotoolkit.org/repositories/openvino_genai/packages/pre-release/2026.5.0.0rc2/openvino_genai_ubuntu26_2026.5.0.0rc2_x86_64.tar.gz | tar -xz &&
sudo mv openvino_genai_ubuntu26_2026.5.0.0rc2_x86_64 /opt/intel/openvino_genai
source /opt/intel/openvino_genai/setupvars.sh
sudo tee /etc/yum.repos.d/openvino.repo >/dev/null <<'EOF'
[OpenVINO]
name=Intel(R) Distribution of OpenVINO
baseurl=https://yum.repos.intel.com/openvino
enabled=1
gpgcheck=1
repo_gpgcheck=1
gpgkey=https://yum.repos.intel.com/intel-gpg-keys/GPG-PUB-KEY-INTEL-SW-PRODUCTS.PUB
EOF
sudo dnf install -y openvino-2026.3.1

NOTE: No Fedora-specific nightly OpenVINO™ GenAI archive is published. These instructions install the official OpenVINO™ Toolkit 2026.3.1 package from Intel’s YUM repository, matching the Fedora 41 Dockerfile.

wget -O- https://storage.openvinotoolkit.org/repositories/openvino_genai/packages/pre-release/2026.5.0.0rc2/openvino_genai_ubuntu24_2026.5.0.0rc2_x86_64.tar.gz | tar -xz &&
sudo mv openvino_genai_ubuntu24_2026.5.0.0rc2_x86_64 /opt/intel/openvino_genai
source /opt/intel/openvino_genai/setupvars.sh

NOTE: No EMT-specific OpenVINO™ GenAI archive is published; the ubuntu24 archive is used instead, matching the “plain” OpenVINO™ Toolkit archive flavor EMT has always relied on for ABI compatibility.

Step 6: Build Deep Learning Streamer#

To build DL Streamer is it recommended to use the provided makefile for ease of use:

make build

Running this command will build any major missing dependencies and then compile DL Streamer itself.

Step 7: Install Deep Learning Streamer (optional)#

After building DL Streamer you can install it on your local system by running:

sudo -E make install

Step 8: Set up environment#

Set up the required environment variables:

export LIBVA_DRIVER_NAME=iHD
export GST_PLUGIN_PATH="/opt/intel/dlstreamer/Release/lib:/opt/intel/dlstreamer/gstreamer/lib/gstreamer-1.0:$HOME/dlstreamer/build/intel64/Release/lib:$HOME/dlstreamer/build/deps/gstreamer-bin/lib/gstreamer-1.0:$GST_PLUGIN_PATH"
export LD_LIBRARY_PATH="/opt/intel/dlstreamer/Release/lib:/opt/intel/dlstreamer/gstreamer/lib:/opt/intel/dlstreamer/opencv/lib:/opt/intel/dlstreamer/rdkafka/lib:$HOME/dlstreamer/build/intel64/Release/lib:$HOME/dlstreamer/build/deps/gstreamer-bin/lib:$HOME/dlstreamer/build/deps/opencv-bin/lib:$HOME/dlstreamer/build/deps/rdkafka-bin/lib:$LD_LIBRARY_PATH"
export LIBVA_DRIVERS_PATH="/usr/lib/x86_64-linux-gnu/dri"
export GST_VA_ALL_DRIVERS="1"
export PATH="/opt/intel/dlstreamer/Release/bin:/opt/intel/dlstreamer/gstreamer/bin:/opt/intel/dlstreamer/opencv/bin:$HOME/dlstreamer/build/intel64/Release/bin:$HOME/dlstreamer/build/deps/gstreamer-bin/bin:$HOME/dlstreamer/build/deps/opencv-bin/bin:$HOME/.local/bin:$HOME/python3venv/bin:$PATH"
export PKG_CONFIG_PATH="/opt/intel/dlstreamer/Release/lib/pkgconfig:/opt/intel/dlstreamer/gstreamer/lib/pkgconfig::$HOME/dlstreamer/build/intel64/Release/lib/pkgconfig:$HOME/dlstreamer/build/deps/gstreamer-bin/lib/pkgconfig:$PKG_CONFIG_PATH"
export GST_PLUGIN_FEATURE_RANK=${GST_PLUGIN_FEATURE_RANK},ximagesink:MAX
export GI_TYPELIB_PATH="/opt/intel/dlstreamer/gstreamer/lib/girepository-1.0:$HOME/dlstreamer/build/src/gst/metadata/:$HOME/dlstreamer/build/deps/gstreamer-bin/lib/girepository-1.0:/usr/lib/x86_64-linux-gnu/girepository-1.0"
export PYTHONPATH="/opt/intel/openvino_genai/python:/opt/intel/dlstreamer/gstreamer/lib/python3/dist-packages:/opt/intel/dlstreamer/python:/opt/intel/dlstreamer/gstreamer/lib/python3/dist-packages:$PYTHONPATH"
export LIBVA_DRIVER_NAME=iHD
export GST_PLUGIN_PATH="/opt/intel/dlstreamer/Release/lib:/opt/intel/dlstreamer/gstreamer/lib/gstreamer-1.0:$HOME/dlstreamer/build/intel64/Release/lib:$HOME/dlstreamer/build/deps/gstreamer-bin/lib/gstreamer-1.0:$GST_PLUGIN_PATH"
export LD_LIBRARY_PATH="/opt/intel/dlstreamer/Release/lib:/opt/intel/dlstreamer/gstreamer/lib:/opt/intel/dlstreamer/opencv/lib:$HOME/dlstreamer/build/intel64/Release/lib:$HOME/dlstreamer/build/deps/gstreamer-bin/lib:$HOME/dlstreamer/build/deps/opencv-bin/lib:$HOME/dlstreamer/build/deps/rdkafka-bin/lib:$LD_LIBRARY_PATH"
export LIBVA_DRIVERS_PATH="/usr/lib64/dri-nonfree"
export GST_VA_ALL_DRIVERS="1"
export PATH="/opt/intel/dlstreamer/Release/bin:/opt/intel/dlstreamer/gstreamer/bin:/opt/intel/dlstreamer/opencv/bin:$HOME/dlstreamer/build/intel64/Release/bin:$HOME/dlstreamer/build/deps/gstreamer-bin/bin:$HOME/dlstreamer/build/deps/opencv-bin/bin:$HOME/.local/bin:$HOME/python3venv/bin:$PATH"
export PKG_CONFIG_PATH="/opt/intel/dlstreamer/Release/lib/pkgconfig:/opt/intel/dlstreamer/gstreamer/lib/pkgconfig::$HOME/dlstreamer/build/intel64/Release/lib/pkgconfig:$HOME/dlstreamer/build/deps/gstreamer-bin/lib/pkgconfig:$PKG_CONFIG_PATH"
export GST_PLUGIN_FEATURE_RANK=${GST_PLUGIN_FEATURE_RANK},ximagesink:MAX
export PYTHONPATH="/opt/intel/openvino_genai/python:/opt/intel/dlstreamer/gstreamer/lib/python3/dist-packages:/opt/intel/dlstreamer/python:/opt/intel/dlstreamer/gstreamer/lib/python3/dist-packages:$PYTHONPATH"

Enable i915 graphics driver in the system:

sudo vim /etc/default/grub
### Extend the GRUB_CMDLINE_LINUX with i915.force_probe=* ###
sudo grub2-mkconfig -o /boot/grub2/grub.cfg "$@"
sudo reboot

After a reboot, before trying the Deep Learning Streamer pipelines, you can export the following environment variables for the current terminal session (temporary solution):

export LIBVA_DRIVER_NAME=iHD
export GST_PLUGIN_PATH="/opt/intel/dlstreamer/Release/lib:/opt/intel/dlstreamer/gstreamer/lib/gstreamer-1.0:$HOME/dlstreamer/build/intel64/Release/lib:$HOME/dlstreamer/build/deps/gstreamer-bin/lib/gstreamer-1.0:$GST_PLUGIN_PATH"
export LD_LIBRARY_PATH="/opt/intel/dlstreamer/Release/lib:/opt/intel/dlstreamer/gstreamer/lib:/opt/intel/dlstreamer/opencv/lib:$HOME/dlstreamer/build/intel64/Release/lib:$HOME/dlstreamer/build/deps/gstreamer-bin/lib:$HOME/dlstreamer/build/deps/opencv-bin/lib:$HOME/dlstreamer/build/deps/rdkafka-bin/lib:$LD_LIBRARY_PATH"
export LIBVA_DRIVERS_PATH="/usr/lib/dri"
export GST_VA_ALL_DRIVERS="1"
export PATH="/opt/intel/dlstreamer/Release/bin:/opt/intel/dlstreamer/gstreamer/bin:/opt/intel/dlstreamer/opencv/bin:$HOME/dlstreamer/build/intel64/Release/bin:$HOME/dlstreamer/build/deps/gstreamer-bin/bin:$HOME/dlstreamer/build/deps/opencv-bin/bin:$HOME/.local/bin:$HOME/python3venv/bin:$PATH"
export PKG_CONFIG_PATH="/opt/intel/dlstreamer/Release/lib/pkgconfig:/opt/intel/dlstreamer/gstreamer/lib/pkgconfig::$HOME/dlstreamer/build/intel64/Release/lib/pkgconfig:$HOME/dlstreamer/build/deps/gstreamer-bin/lib/pkgconfig:$PKG_CONFIG_PATH"
export GST_PLUGIN_FEATURE_RANK=${GST_PLUGIN_FEATURE_RANK},ximagesink:MAX
export PYTHONPATH="/opt/intel/openvino_genai/python:/opt/intel/dlstreamer/gstreamer/lib/python3/dist-packages:/opt/intel/dlstreamer/python:/opt/intel/dlstreamer/gstreamer/lib/python3/dist-packages:$PYTHONPATH"

NOTE: For a permanent solution, open \~/.bashrc and add the variables above to set up Linux to use them for every terminal session.

Step 9: Install Python dependencies (optional)#

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

NOTE: This step requires the ~/python3venv virtual environment and the uv tool set up in Step 3. If you skipped that step, install uv and create the virtual environment first.

NOTE: Install PyGObject into the activated ~/python3venv from Step 3 so it is built for that environment’s Python ABI. Ubuntu 22.04 uses uv-managed Python, which includes its own development headers. Ubuntu 24.04 and 26.04 use the system Python, so install its matching python3-dev.

sudo apt-get update
sudo apt-get install -y -q --no-install-recommends gcc cmake python-gi-dev python3-dev \
    libglib2.0-dev libcairo2-dev libopencv-objdetect-dev libopencv-photo-dev libopencv-stitching-dev libopencv-video-dev \
    libopencv-calib3d-dev libopencv-core-dev libopencv-dnn-dev libgirepository1.0-dev

source ~/python3venv/bin/activate
cd ~/dlstreamer
uv pip install -r requirements.txt
sudo apt-get update
sudo apt-get install -y -q --no-install-recommends gcc cmake python-gi-dev \
    libglib2.0-dev libcairo2-dev libopencv-objdetect-dev libopencv-photo-dev libopencv-stitching-dev libopencv-video-dev \
    libopencv-calib3d-dev libopencv-core-dev libopencv-dnn-dev libgirepository1.0-dev

source ~/python3venv/bin/activate
cd ~/dlstreamer
uv pip install -r requirements.txt
sudo apt-get update
sudo apt-get install -y -q --no-install-recommends gcc cmake python-gi-dev python3-dev \
    libglib2.0-dev libcairo2-dev libopencv-objdetect-dev libopencv-photo-dev libopencv-stitching-dev libopencv-video-dev \
    libopencv-calib3d-dev libopencv-core-dev libopencv-dnn-dev libgirepository1.0-dev libgirepository-2.0-dev

source ~/python3venv/bin/activate
cd ~/dlstreamer
uv pip install -r requirements.txt

NOTE: Fedora 41 / EMT users: install the dnf equivalents of the packages above (see Step 2) before running uv pip install.

(Optional) Install DL Streamer ONVIF Python package#

If you want to use the dlstreamer.onvif in your own Python code, install the intel-dlstreamer wheel on top of the DL Streamer. See ONVIF sample for more details of ONVIF support.

The easiest way is to download and install the wheel directly from the GitHub Release:

uv pip install https://github.com/open-edge-platform/dlstreamer/releases/download/v2026.2.0/intel_dlstreamer-2026.2.0-py3-none-any.whl

Alternatively, build the wheel from the cloned sources:

uv build --wheel ~/dlstreamer/python
uv pip install ~/dlstreamer/python/dist/intel_dlstreamer-*.whl

This installs:

  • dlstreamer.onvif — ONVIF camera discovery and pipeline integration

  • All required Python dependencies (numpy, onvif-zeep, urllib3) automatically

After installation, no PYTHONPATH configuration is needed:

from dlstreamer.onvif import DlsOnvifDiscoveryEngine, discover_onvif_cameras

NOTE (advanced): As an alternative to pip, dlstreamer.onvif is also accessible via PYTHONPATH set in Step 8. This approach requires manual installation of the dependencies declared in python/pyproject.toml.


Intel, the Intel logo, OpenVINO, and the OpenVINO logo are trademarks of Intel Corporation or its subsidiaries.