This was a machine learning project that used both k-means clustering as well as a neural networks to attempt to predict whether a given tomato leaf image is diseased or not. The initial data set was found on Kaggle @ https://www.kaggle.com/datasets/ashishmotwani/tomato
The goal was to find which technique worked better. Ultimately, I found neither of them to work well, but I learned a lot about the computational limits of the desktop I was using to train a neural net (as I was reaching the peak). A neural network approach did seem to be best between both the methods, but I got stuck at little more than 50% accuracy (which, considering it was categorizing between 10+ categories, is not that bad of a start). I learned quite a bit about k-means, and how it detects clusters that I might not think of.
Attached is the code I used to train the k-means as well as the neural nets. I also attached the slides used for the presentation on this work.