project url: https://www.kaggle.com/c/plant-pathology-2020-fgvc7/overview

Arpit's colab url: https://colab.research.google.com/drive/1XAwCLRV947r3K3eZnoqWfiO0I5ySLMNq?usp=sharing#scrollTo=OVkwHdV28UaY

Objectives:

1. Classify if healthy or not
2. If not, figure out which disease
3. Rare classes -> unbalanced dataset
4. Angle, light, shade, age -> data augmentation/ pre-processing
5. Additional features -> extra features to detect

To-do:

1. Successfully classify between healthy or not i.e. look for higher accuracy
2. Take step 2 results and build model for detecting which disease -> higher accuracy the better

Quick initial observation:

1. If healthy, scab=0, rust=0, multiple diseases=0