Previous attempts at classifying plant diseases from images have yield positive results. This Kaggle challenge addresses the issue of visual variations in symptoms by providing a much richer training dataset which exposes a ML algorithm to variations in natural and image capturing environments including but not limited to leaf color and leaf morphology, the age of infected tissues, non-uniform image background, and different light illumination during imaging.
To address this image classification problem, we can further augment the training dataset by applying a number of image modifcations, such as playing on brigthness, angle, noise, and contrast. The next step is building a neural network and deciding whether to benefit from transfer learning or training a new model from scratch.