Automated Rice Plant Nutrient Deficiency Classification using Manta Ray Foraging Optimization with Denoising Autoencoder

Abstract

Agriculture is a major element of the world's economic and food supplies that is under strain
as an outcome of the enormous increase in population. Digital agriculture is a novel scientific
area that supports agricultural production. Most the several features that stimulate agricultural
produce is the lack of soil nutrients. Micronutrient acquisition by crops is assumed to
progress suitable approaches for preventing a crop deficiency. With the accessibility of higher
computational power, it can be possible for enhancing the efficiency of a predicting approach
with utilize of advanced Deep Learning (DL) systems. This study introduces an Automated
Rice Plant Nutrient Deficiency Classification using Manta ray Foraging Optimization with
Denoising Autoencoder (NDC-MRFODAE) technique. The primary goal of the NDCMRFODAE
technique lies in the recognition of N, P, and K nutrients in rice plants. To
achieve this, the NDC-MRFODAE technique uses Wiener filtering (WF) approach to get rid
of the noise in the input rice plant image. Followed, neural architectural search (NASNet)
model is used for feature extraction purposes and the hyperparameter tuning process is
carried out by the MRFO algorithm. For classification process, DAE classifier is exploited in
this work, which helps in accomplishing improved nutrient deficiency classification. The
experimental assessment of the NDC-MRFODAE technique is validated on benchmark
dataset for nutrient deficiency recognition. The obtained values highlighted the enhanced
performance of the NDC-MRFODAE approach over recent state of approaches.