Deep Learning Based Sugarcane Downy Mildew Disease Detection Using CNN-LSTM Ensemble Model for Severity Level Classification
Nikhil Dhawan, Vinay Kukreja, Rishabh Sharma, Satvik Vats, Aditya Verma
Abstract
The fungus known as sugarcane downy mildew is extremely destructive and represents a substantial risk to sugarcane output all over the world. To effectively manage the disease and safeguard crops, early and precise diagnosis of the severity levels caused by downy mildew is essential. In this study, we present a novel method for detecting sugarcane downy mildew disease based on severity levels using a CNN-LSTM ensemble model. The ensemble model combines the spatial feature extraction skills of Convolutional Neural Networks (CNN) with the temporal modeling abilities of Long Short-Term Memory (LSTM) networks. For model training and assessment, a dataset consisting of sugarcane leaf pictures that have been afflicted by downy mildew and labeled with severity levels ranging from 1 to 5 is employed. The results of the experiments show that the suggested ensemble model is successful, as it achieves high levels of accuracy, precision, recall, and F1 score while attempting to forecast the severity levels of downy mildew. The capacity of the model to effectively categorize severity levels throughout the dataset is demonstrated by the overall accuracy of the model as 94.16%. The CNN-LSTM ensemble model that was provided provides a potential solution for the automated identification of the illness known as sugarcane downy mildew. The suggested method enables prompt interventions and optimizes disease management practices. The proposed study contributes to the growing field of agricultural informatics and helps promote environmentally responsible methods of growing sugarcane.