Leveraging three-tier deep learning model for environmental cleaner plants production

Area of Science:

• Agricultural Technology
• Computational Science
• Internet of Things (IoT)

Background:

• Global population growth to 9 billion by 2050 requires a 70% increase in food production.
• Challenges include resource scarcity, climate change, and pandemics, necessitating computational forecasting.
• Plant diseases pose significant threats from seed to growth stages.

Purpose of the Study:

• To develop an early plant disease diagnosis framework using fog computing and edge environments.
• To evaluate the effectiveness of pre-trained Convolutional Neural Network (CNN) architectures as feature extractors.
• To improve the accuracy of plant disease identification for enhanced agricultural output.

Main Methods:

• Utilized IoT sensors for data collection in fog and edge computing environments.
• Employed pre-trained CNN models (AlexNet) as feature extractors.
• Applied a revised Grey Wolf Optimization (GWO) algorithm for feature selection and trained an SVM classifier.

Main Results:

• The proposed model achieved an average accuracy of 93.84% across ten diverse plant datasets.
• This surpasses the accuracy of standard AlexNet (85.49%), GoogleNet (87.89%), and SVM (87.04%).
• The GWO algorithm effectively optimized feature selection for improved classification.

Conclusions:

• The proposed framework offers a highly accurate solution for early plant disease diagnosis.
• This contributes to more reliable agricultural forecasting and increased food production.
• The integration of IoT, fog computing, and optimized machine learning shows significant promise in agri-technology.