Image Processing based Plant Disease Detection and Classification

Abstract

Generally, it has been observed that due to lack of proper knowledge of disease intensity, the farmer is not able to use the pesticide in proper quantity to treat the diseases. The use of pesticide mostly becomes more than necessary, due to which there is not only a loss of money, but also it causes soil and environmental pollution. If diseases severity-wise labelled data sets are available, it can be used to develop pesticide recommendation systems. Images with least infection severity can be used to train and validate a DL model to capture the plant diseases at very initial stage. Classification techniques may be viewed as variations of detection systems; however, instead of attempting to identify only one particular illness among several diseases, classification methods detect and name the diseases harming the plant. This presents various classification and plant disease detection methods based on image processing with results.