SAILFISH OPTIMIZATION ALGORITHM WITH DEEP CONVOLUTIONAL NEURAL NETWORK FOR NUTRIENT DEFICIENCY DETECTION IN RICE PLANTS

Abstract

Computer vision (CV)-based automation has gained popularity in monitoring and detecting plants’ nutrient deficiency. The prediction method formulated by several authors can be utilized in an embedded mechanism, considering the accessibility of computational resources. Yet, the huge popularity of smartphone technology paved the way for common agronomists to have access to high computational resources. Therefore, this study presents a Sailfish Optimization with deep learning for Nutrient Deficiency Detection (SFODL-NDD) in Rice Plants. The presented SFODL-NDD technique focuses on the identification of nutrients using DL and computer vision approaches. Initially, the SFODL-NDD technique employs median filtering (MF) approach for noise elimination process. Next, the SFODL-NDD technique uses Xception model for feature extraction with SFO algorithm as hyperparameter optimizer. At last, the nutrient deficiencies can be recognized by extreme gradient boosting (XGBoost) classifier. The experimental evaluation of the SFODL-NDD technique is tested using rice plant dataset from Kaggle repository. The experimental outcomes stated the better superior performance of the SFODL-NDD technique over other approaches in terms of different measures.