Installing Darknet on Windows 10 without OpenCV and CUDA (no GPU on own device): issues

Question

I have been trying to install Darknet on my own device (which runs on Windows 10) unsuccessfully. I am following the rough guide here: https://pjreddie.com/darknet/install/. However, it only contains the steps for a Linux configuration, so I have been checking out posts and videos about doing it on Windows.

Has anyone come across a relatively straightforward guide or YouTube video about how to install Darknet without the need to have a local GPU (as a dependency, CUDA requires GPU) that you have found particularly useful? I have been following a few (https://www.youtube.com/watch?v=saDipJR14Lc was pretty clear), but they all go through the CUDA installation as a must (where CUDA should be an optional dependency).

Many thanks in advance!

Answer 1

You can use this GitHub repository for installing darknet. Refer to this section for installing it on Windows 10.

There is a detailed description provided on how to go about installing it on Windows 10 without GPU and OpenCV support

Answer 2

In the end, I was able to successfully install Darknet using the following 3 lines of code in Powershell (after installing the required dependencies, with ninja this time around, as well as cloning the Alexey AB repository):

cd darknet
git pull 
.\build.ps1 -EnableCUDA -EnableCUDNN -EnableOPENCV -EnableOPENCV_CUDA -UseVCPKG

The discussions on the Discord forum were extremely helpful: https://discord.com/channels/741676058666860635/741676058666860639