IoT enhanced metaheuristics with deep transfer learning based robust crop pest recognition and classification

Abstract

This study addresses the pressing need to enhance food production and quality by tackling the challenge of crop pest recognition. Traditional methods for identifying crop pests are time-consuming and reliant on highly trained experts. To overcome these limitations, this research introduces the Enhanced Metaheuristics with Deep Transfer Learning-based Robust Crop Pest Recognition and Classification (EMDTL-CPRC) model. The EMDTL-CPRC model incorporates several innovative elements, including contrast enhancement to improve image quality, the integration of the Whale Optimization Algorithm (WOA) with the Residual Network (ResNet) approach as a feature extractor, and the utilization of the Long Short-Term Memory (LSTM) method for pest recognition. Furthermore, to enhance the efficacy of the LSTM approach, a Chaotic Salp Swarm Algorithm (CSSA) is incorporated. The simulation results, based on a comprehensive crop pest dataset, demonstrate the superior performance of the EMDTL-CPRC model in crop pest classification. By enhancing pest recognition accuracy and streamlining the process, this research contributes to the overall goal of improving agricultural crop productivity and ensuring the supply of healthy food.