Pepper Leaf Diseases Detection Using Hybrid Techniques with Deep Learning

Abstract

Plant disease detection in agriculture is a significant challenge, with potential benefits in monitoring crop fields and identifying symptoms on plant leaves. However, most systems underperform due to varying leaf disease patterns, limiting detection precision. Automating the recognition of leaf diseases can boost farming production. To solve this problem, we proposed a novel state-of-the-art hybrid deep learning model. Our objectives are to construct a better pepper leaf disease detection method and model using hybrid deep learning techniques based on CNN-CAE and CNN-HOG models that have been trained, validated, and tested. The performance of each model for detecting pepper leaf disease was assessed. The hybrid CNN-HOG and CNN-CAE model was giving us good and promising results such as accuracy, precision, recall, Fl-Score performance metrics. The findings suggest that hybrid deep learning can detect pepper leaf illness better. Farmers and agriculture specialists can identify and manage pepper leaf illnesses faster with such approaches, improving pepper output and sustainability.