Plant Disease Identification and Suggestion of Remedial Measures using Machine Learning

Abstract

Plants are an important source of energy for all organisms on earth. But plant diseases act as a hindrance for effective consumption of plant products and also adversely affect the life of crops. When the farmers diagnose diseases manually, lot of difficulties arise due of the lack of knowledge and unavailability of professionals. It also requires much time in manually identifying and classifying crop diseases. In this context, a model is proposed for identifying plant diseases and to suggest remedial measures. Here a transfer learning based CNN model is implemented using VGG16 and ResNet50. The dataset used consists of 34824 training images and 8767 testing images of thirty-eight output classifications including 26 crop diseases found in fourteen crops. The VGG16 model shown 99.1 percentage accuracy and ResNet50 exhibited 99.3 percentage accuracy with considerable reduction of computation time than VGG16.