I would like to build a small style transfer neural network, via darknet (Open Source Neural Networks in C), under a Debian 11 32-bit system, for experimental and personal use. What I intend is to test an image and another identical to it, but to which I have applied filters. I have no intention of using pre-trained models and I don't know if they exist for this software. Given my system architecture (32 bit), I can't use Pytorch or Tensorflow with Python. Would it be possible to do it with darknet?. If so, what would be the first steps to take?. Thank you very much in advance and a warm greeting to the community.
Asked
Active
Viewed 7 times
0
-
So far, the steps taken: Step 1: Darknet installation. In my case I have chosen to install, from Synaptic, the package: "darknet" Your location in the system: /usr/bin/darknet In the "usr/share/darknet/cfg" directory, you will find the configuration files: Among them, it is worth noting "writing.cfg". Step 2 – Download the pre-trained YOLOv3 models. https://pjreddie.com/media/files/yolov3.weights Move to: "/usr/bin/darknet". Step 3: Create a configuration file. You can use the example file mentioned above: /usr/share/darknet/cfg/writing.cfg – veron.schenone Sep 01 '23 at 19:51