My lesson 1 homework was to try entering the ‘foliar disease recoginition’ kaggle competition. I thought this would be cool since I grow a lot of apple trees and deal with diseases like these, plus many more.
I’m happy to have done this as my assignment because it taught me how to load training data using a different method than the lesson. I got to use ImageDataBunch.from_csv(), which took a little work.
By way of making some mistakes, I also learned the importance of setting size and bs arguments (I ran out of memory).
However, my error rate is very high. I’m not sure if that’s because this data is harder for the algorithm or if I’m doing something wrong. Any insight into what I did wrong and how I could do better is much appreciated!