Multi-Plant Leaf Disease Identification Using Vein-Related Symptom Analysis

Abstract

A promising solution to enhance yield efficiency is Plant Leaf Disease Detection (PLDD). Nevertheless, none of the traditional frameworks focused on the vein-related symptoms during PLDD.  Therefore, this paper proposes a robust MER-DS-EfficientNet and EZWS-based vein-related symptom analysis and multi-PLDD. Initially, the plant Leaf Disease (LD) datasets are collected and then pre-processed. Then, the background region is eliminated, followed by density heat map generation. Afterward, the feature extraction and feature reduction are done. Here, to identify the type of plant leaf, the proposed MER-DS-EfficientNet is utilized. Then, to recognize the healthy and disease-affected leaves, the Color Component Analysis (CCA) is done. Next, the disease region is segmented, followed by feature extraction and feature reduction. Also, via the hessian matrix, the Vein Region (VR) is fragmented. Then, from the vein factors, the VER is computed. Lastly, the proposed MER-DS-EfficientNet significantly classifies the multi-plant LDs. The proposed method performed better, with 98.9942% accuracy ac tincidunt vitae semper.