Disease Recognition of Crops using ResNet and MDFC-ResNet

Abstract

In the field of agriculture, recognizing a crop disease is often a crucial task. Identification of crop disease directly impacts the treatment strategy for the disease. Without proper identification, the disease control measures could be waste of time and cost if an inappropriate approach is chosen. Ineffective measures have a negative impact on crop yield and food safety in addition to ineffectively treating diseases. Deep Convolutional Neural Networks (DCNN) with leaf image classification is used to identify crop diseases. The model identifies crop diseases and sends the results to the users through a web application. Residual neural network and multidimensional feature compensation residual network (MDFC-ResNet) model is used for recognizing the disease and their performance is compared. The models identify the species of crop along with the disease of crop. The models are capable of distinguishing diseased leaves and healthy leaves from the environment.