In this section, we provide instructions how to install OTVision on the most common Operating Systems.
Before installing OTVision, make sure your system meets all requirements.
We provide install scripts for the most common operating systems.
Inside the unzipped folder open the
install.cmd and wait until the
installation of the dependencies is complete.
In a terminal, navigate to the OTVision folder and run the installer.
The installation of the dependencies could take a moment.
What is installed here?
install script will create and activate a virtual environment (venv)
and install the Python packages specified in the requirements.txt via pip
from the Python Package Index.
Nvidia CUDA (optional)¶
If you have a
Windows or Linux PC with a Nvidia graphics card
and already installed CUDA,
you chose the release with the suffix
-cuda. It contains the requirements to use CUDA.
If you want to contribute to OTVision and use CUDA, you have to perform additional
steps in your Terminal/Command Prompt:
Check CUDA version¶
Check if CUDA is recognized and available.
Navigate to the OTVision root directory.
Where is the OTVision root directory?
It's the folder you downloaded und unzipped.
Maybe your OTVision root directory is called
OTVision-main after unzipping,
if you downloaded it from Github. This is the correct directory.
Inside the OTVision root directory, there is another directory called
(this child directory is the wrong directory).
Activate virtual environment¶
Activate the virtual environment that was created by running the installation scripts.
Open a Command Prompt an run:
The virtual environment should be activated, indicated by the
in braces in front of your current working directory in the terminal.
Install torch and torchvision for CUDA¶
If you downloaded a
-cuda release, you are good to go, if the CUDA version
requirements.txt matches your system.
To install another version you can do so follwing the PyTorch documentation:
Depending on your operating system (Windows or Linux) and your CUDA version you can select, copy and run the install command from the PyTorch site under "INSTALL PYTORCH" (choose Build="Stable", Package="pip" and Language="Python").
E.g., for CUDA 11.6 and the latest stable PyTorch Build, the command is:
pip3 install torch torchvision --extra-index-url https://download.pytorch.org/whl/cu116
If you encounter problems¶
Maybe you also have to install Microsoft Visual C++ 14.0 or greater from the Visual Studio Build Tools.
In case of further problems please open an issue in the OTVision repository on GitHub or contact us. We are happy to know about your experience.
We welcome code contributions (e.g., fixing bugs or adding features) from others by forking the repository and creating a pull request. Please check the contribute section of this documentation first.
If you want to contribute code, additional requirements should be installed
in the virtual environment.
Clone the OTVision repository
from GitHub. Run the
install_dev.sh in your OTVision folder
and wait until the installation of the dependencies is complete.