An Enhanced CNN-based ELM Classification for Disease Prediction in the Rice Crop

Abstract

To meet the demands of a constantly expanding population, intensive farming is becoming more popular in the modern day. This strategy, meanwhile, increases the possibility of a wider range of plant illnesses. By reducing crop productivity in terms of both quantity and quality, these infections represent a threat to food production and ultimately result in a fall in the economy. Fortunately, new opportunities for early diagnosis of such epidemics have emerged because of technological improvements, which are advantageous for society as a whole. The difficulties created by technology and bio-mutations create a potential for additional breakthroughs, notwithstanding the significant contributions made by researchers in the field of agricultural disease diagnosis. The suggested framework comprises three key phases: preprocessing, feature extraction, and the classification of leaf diseases. To optimize computational resources and memory utilization, the input image undergoes pre-processing as a preliminary step. Afterward, a Convolutional Neural Network (CNN) is utilized on an extensive dataset of labeled images to capture pertinent features for the diagnosis of rice leaf diseases. The suggested model utilizes an Efficient Selective Pruning of Hidden Nodes (ELM) classifier based on the RBF kernel to classify the input data.