1 - Install YOLO11

1.2.1 - Before installing YOLO

It is recommended to run trackRai on a machine equipped with an NVIDIA graphics card and the CUDA toolkit. If your computer is equipped with an NVIDIA graphics card, you can find instructions to install CUDA here:

• For Windows computers: https://docs.nvidia.com/cuda/cuda-installation-guide-microsoft-windows/
• For Linux computers: https://docs.nvidia.com/cuda/cuda-installation-guide-linux/

If your computer is not equipped with an NVIDIA graphics card or if the CUDA toolkit is not installed on it, you can still use trackRai but the training of the YOLO model may take a very long time (several hours).

Note: at the time of writing, YOLO (and therefore, trackRai) is compatible with CUDA versions 11.8, 12.4, and 12.6.

1.2.2 - Installing YOLO

Once trackRai is installed, you will need to install YOLO on your machine using the helper function install_yolo() provided with trackRai.

library(trackRai)
install_yolo()

The installer will prompt you with a series of questions. Answer “Yes” every time if you want YOLO and its dependencies to be installed on your machine. Answering “No” may stop the installation process and trackRai may not be usable. You may also need to restart your R session before being able to use the apps provided with trackRai.

Note: by default, install_yolo() will attempt to install Python 3.12.5 on your system if it is not already present. If errors happen during the installation of Python, or if you prefer installing another version, you can do so using the python_version parameter of install_yolo(). YOLO is currently compatible with Python 3.8.0 to 3.12.8, so any of these versions should work.

Note: on a Windows computer, install_yolo() will attempt to automatically detect the version of CUDA that is installed on the machine. You can override this using the cuda_win_version parameter.