Download¶

Python¶

pipx install dvr-scan[opencv]

python3 -m pip install dvr-scan[opencv]

DVR-Scan requires Python 3.9 or higher to run, and works on Windows, Linux, and OSX. pipx is recommended for installing DVR-Scan, however installing via pip or from source is also supported.

Linux users may need to install the python3-tk package (e.g. sudo apt install python3-tk) to run the region editor.

Windows¶

The installer is recommended for most users. Windows builds include all required dependencies to run DVR-Scan. Only 64-bit builds are available.

Servers¶

For headless systems that do not require the UI, you can install dvr-scan[opencv-headless]. This will make sure that the headless version of OpenCV is installed, which avoids any dependencies on X11 libraries or any other GUI components. This allows DVR-Scan to run with less dependencies, and can result in smaller Docker images.

Source¶

The source code for DVR-Scan is available on Github. It can be run directly from source (python -m dvr_scan), or built locally (python -m build).

CUDA® Support¶

DVR-Scan works with CUDA graphics cards if you are using the Python distribution, and you have a CUDA-enabled verison of the opencv-python package.

It is recommended to build and install OpenCV with the CUDA module enabled (there are various tutorials online for how to do this). There is also an unofficial pre-built Python wheel maintained by James Bowley which can be downloaded and installed via pip.

When available, you should see cv2.cuda: Installed under the features list when running dvr-scan --version. Make sure to set -b MOG2_CUDA when running DVR-Scan (e.g. dvr-scan -i video.mp4 -b MOG2_CUDA). In the UI, under this can be found under Settings -> Motion -> Subtractor.

Code Signing Policy

Free code signing provided by SignPath.io, certificate by SignPath Foundation.