Created on 19th December 2021
•
Plants are susceptible to disease for a variety of reasons, including fertilisers, cultural practices, environmental conditions, and so on. These diseases reduce agricultural yield and, as a result, the economy that relies on it. Any approach or method for overcoming this challenge and receiving a warning before the plants become infected would help farmers cultivate crops or plants more efficiently in terms of both quality and quantity. As a result, disease detection in plants is critical in agriculture.
Ref: https://www.tensorflow.org/datasets/catalog/plant_village
The PlantVillage dataset consists of 54303 healthy and unhealthy leaf images divided into 38 categories by species and disease.
The model can be migrated to a cloud environment and a web app can me made to use this in fields by farmers who would benefit a lot from this.
We had trouble dealing with this large dataset which took much computational power so we migrated from the local Jupyter Notebook to Google Colab which increased our speed which allowed us to research a few algorithms. ResNet seems to work better than VGG or Inception.
Technologies used