Install Guide Windows (Preview)#

Building Intel® Deep Learning Streamer (Intel® DL Streamer) Pipeline Framework from the source code provided in this repository may be performed either on host system or as a Docker image.

Option 1: Compile Intel DL Streamer Pipeline Framework from sources on host system#


These instructions are verified on Windows 10 Version 1909 and Windows 11 Version 10.0.22000.

Step 1: Clone this repository#

First, clone this repository into folder %USERPROFILE%\intel\dlstreamer_gst:

md %USERPROFILE%\intel
git clone %USERPROFILE%\intel\dlstreamer_gst
cd %USERPROFILE%\intel\dlstreamer_gst

Step 2: Install dependencies#

Install Python 3.7 for Windows:

Install CMake 3.19 for Windows:

Install Visual Studio 2019* with C++ build tools:

Install pkg-config using one of the options:

Install GStreamer 1.18.4 runtime and development packages:


If you want to use web camera as a video source then mark ‘Gstreamer 1.0 plugins for capture’ for installation

Setup environment variables for GStreamer:

set PATH=%PATH%;C:\gstreamer\1.0\msvc_x86_64\bin
set PKG_CONFIG_PATH=%PKG_CONFIG_PATH%;C:\gstreamer\1.0\msvc_x86_64\lib\pkgconfig

Step 3: Download and install Intel® Distribution of OpenVINO™ Toolkit#

Please register to download the latest version of OpenVINO™ Toolkit for Windows*.

For the installation recommended to follow the OpenVINO™ Toolkit Installation Guide.

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

@REM Install OpenCV
powershell.exe -ExecutionPolicy Bypass -File "C:\Program Files (x86)\Intel\openvino_2022\extras\scripts\download_opencv.ps1"

@REM Install Open Model Zoo tools
pip install openvino-dev[onnx]

@REM Setup environment
"C:\Program Files (x86)\Intel\openvino_2022\setupvars.bat"

Step 4: Build Intel® DL Streamer Pipeline Framework#

With all dependencies installed, proceed to build Pipeline Framework:

@REM OpenVINO™ Toolkit environment
"C:\Program Files (x86)\Intel\openvino_2022\setupvars.bat"

@REM cmake
md %USERPROFILE%\intel\dlstreamer_gst\build
cd %USERPROFILE%\intel\dlstreamer_gst\build
cmake ..

@REM build
cmake --build . --config Release

@REM Setup env variables with script

@REM Verify GStreamer and Intel® DL Streamer Pipeline Framework plugins are now available
gst-inspect-1.0 gvadetect

Pipeline Framework is now ready to use!

You can find samples in folder %USERPROFILE%\intel\dlstreamer_gst\samples. Before using the samples, run the script download_models.bat (located in samples folder) to download the models required for samples.


The environment variables are removed when you close the command line. As an option, you can set them in system “Environment Variables” settings.

Option 2: Compile Intel® DL Streamer Pipeline Framework as Docker image#

Step 1: Install Docker Desktop (if not installed)#

Refer to Docker installation documentation


Default setup uses WSL2 for Docker daemon. To support Windows build steps below, choose Switch to Windows containers... from the Docker Desktop menu in the system tray.

Step 2: Clone this repository#

Clone this repository into folder %USERPROFILE%\intel\dlstreamer_gst:

md %USERPROFILE%\intel
git clone %USERPROFILE%\intel\dlstreamer_gst
cd %USERPROFILE%\intel\dlstreamer_gst

Step 3: Build Docker image#

Run the following command to build Docker image:

cd %USERPROFILE%\intel\dlstreamer_gst\docker\source\windows
@REM provide a link to the OpenVINO™ Toolkit development package as a first argument (by default link to 2022.1 package)
@REM provide a specific image tag (by default latest)
@REM if you want to use a specific Windows image then provide its full name (by default
build_docker_image.bat <link> [tag] [base image]


You can provide a link to OpenVINO™ Toolkit development package from public storage. For example,


If you are behind a corporate proxy you may need to update Docker configuration and Powershell steps in the Dockerfile. For specific guidance on vs_buildtools.exe, refer to Visual Studio Build Tools.

The build process may take significant time and should finally create Docker image dlstreamer:<tag> (by default latest). Validate Docker image with command:

docker images | findstr dlstreamer

This command will return a line with image dlstreamer:<tag> description. If the image is absent in your output, please repeat all the steps above.

Step 4: Run Docker image#

Run container:

docker run -it ^
-v %USERPROFILE%\intel\dl_streamer\models:C:\intel\dl_streamer\models ^
-e MODELS_PATH=C:\intel\dl_streamer\models ^

Here is the additional information and the meaning of some options in the Docker run command:

  • -v instances are needed to map host system directories inside Docker container.

  • -e instances set Docker container environment variables. The samples need some of them set in order to operate correctly.

  • Volume provided for models folder will be used to download and store the models. Environment variable MODELS_PATH is responsible for it.

Container will start with development environment for Visual Studio* and OpenVINO™ Toolkit and open PowerShell* in samples folder.


  • gvatrack and gvapython elements are not available

  • gvametapublish works only with filesystem and console output

  • Docker image contains only basic GStreamer plugins. Use gst-inspect-1.0 command for a full list of available plugins.

  • Refer to OpenVINO™ Toolkit for Windows* for requirements to run inference on GPU in Docker

Next Steps#

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