Comprehensive Deep Learning Approach for Identifying Plant Nutrient Deficiency, Diseases and Pests

Abstract

The core of the Indian economy is agriculture, facilitating the expansion of the service or industrial sectors. The quantity and standard of agricultural goods are seriously threatened by plant diseases. For agricultural productivity, early symptom detection and precise illness classification are essential. This paper underscores the transition from traditional methods, such as image processing and deep learning due to their improved efficiency and accuracy. The primary focus lies in classifying plant diseases using various Deep Learning architectures to achieve higher accuracy. The study will also provide predicting plant disease along with Nutrient Deficiency and pests identification. The final goal will be detecting all these three outcome using single model with high accuracy so no need to execute multiple times. Dataset used in this research is OLID I(Open Leaf Image Dataset). Models are trained with augmented images as well as original dataset. It is found that EfficientNetV2B0 model outperforms some state-of-the-art convolutional neural networks that include EfficientNetV2B3, VGG16 and custom CNN model while running with augmented dataset. Which achieves accuracy with 85.38% and f1-score with 85.08. © 2024 IEEE.