ABSTRACT

It is observed that a lot of crop gets wasted every year due to the illness spreading (reasons can be pests, bacterial or fungal infections, etc.) in the farms. Farmers face huge losses because of the lack of a handy and economical technology that can detect such diseases in the initial stages of occurrence. Model PlaNet (PLANT LEAF ANALYSER NETWORK) can prove a remedy to this issue by developing a prototype that recognizes the same at an early stage and thence providing them a way to cure it using appropriate pesticides, insecticides, etc.

CONCEPT

A deployable system is built for plants’ disease detection by training a Convolutional Neural Network model (RESNET) that traces a particular disease in a particular plant species. It detects the plant disease by scanning the leaves of the species as it is trained for those illness occurring on the leaves.

Workflow

• A dataset containing healthy as well as infected leaf samples of various species is collected and the model is trained for these using deep nets.
• At the time of detection, a new sample is taken and processed in the same manner and then mapped first for the species’ classification and then for the corresponding disease and its remedies.