AI based hybrid CNN-LSTM model for crop disease prediction: An ML advent for rice crop
Sonal Jain, Dharavath Ramesh
Abstract
Plant disease is an extreme challenge in gaining appropriate yield and crop quality. Therefore, a pest forewarning system is advantageous in early disease prediction and controlling it by practicing suitable measures. This paper presents a pest prediction and classification model for yellow stem border (YSB) disease in rice plants. An Artificial Intelligent based prediction model developed considering historical pest and weather data of various regions of India. The proposed model is named as hybrid CNN-LSTM that combines the advantage of convolution neural network (CNN) and long short term memory network (LSTM). It is a region-specific prediction model that predicts one-month pest data based on past three-months weather and pest data. The performance of the proposed model is compared with CNN and LSTM networks. This shows the enhancement in performance while using hybrid CNN-LSTM. On the other hand, this paper also presents a generalized classification model by combining the datasets of all regions. The model predicts the disease severity for the next day based on weather and preceding day pest data. The error correcting output code (ECOC) method with SVM classifier is used for the classification of disease severity.