hi @mpillar , thanks for your project, I am learning the related knowledge recently and found your project really interesting. I am able to run the program now but still confused by some parameters and how this program do the training.
If I run the program ./run.sh fdet mouth ./sample/face1.jpeg 1, what exactly 1 (emotion_type) means? Does it mean, for this input picture, I define this emotion a "1"?
I am trying to understand how the code works to recognize images. I have many questions but the main one at the moment is what are these arguments for:
What exactly do they do when it comes to kohonen or backprop?
Edit: I think i understood what they are for, the problem still remains though. Every time i run the train set i get completely different results. Same train set, same test set, same arguments -> COMPLETELY different results.