PLANT DISEASE DETECTION AND PREVENTION USING DEEP LEARNING

Abstract

Identifying and managing diseases in plants is a critical area of research in agriculture, aiming to ensure optimal crop yield and minimize losses. With India’s large population and the increasing demand for food production, it becomes imperative to address crop diseases effectively and efficiently. Detecting diseases in crops allows for timely interven tion and preventive measures to protect agricultural resources and meet the growing food requirements. Traditionally, disease identification in plants has relied on the expertise of trained professionals who visually inspect crops or conduct chemical tests. However, these methods can be time-consuming, costly, and dependent on the availability of skilled personnel. To overcome these limitations, an automated system that can accurately de tect plant diseases by analyzing plant leaves can revolutionize disease management in crops. By leveraging deep learning models, we can classify leaves based on the specific diseases they exhibit, enabling targeted and timely interventions. In this paper, we focus on detecting diseases in cassava plants, which are widely consumed, including their leaves. The objective of this study is to improve the accuracy and efficiency of disease detection compared to previously proposed models. Deep learning models have shown superior per formance in plant disease detection compared to traditional machine learning methods, prompting their adoption in this study. Specifically, we employ the EfficientNet-B0 archi tecture, which is renowned for its classification capabilities, speed, and scalability in terms of width, depth, and resolution. To ensure robust evaluation and model performance, we utilize k-fold cross-validation, a technique that divides the dataset into k subsets for train ing and validation. This approach helps to assess the model’s generalizability and ensures reliable accuracy measurements. The results obtained from our study indicate a significant advancement in cassava disease detection. Our proposed model achieves an impressive accuracy of 96.68% on the dataset collected from Kaggle, demonstrating the model’s ef fectiveness in identifying diseases in cassava plants. The high accuracy rate achieved by iv the EfficientNet-B0 model provides promising prospects for real-world applications and the potential to enhance disease management practices in cassava cultivation. By leveraging the power of deep learning and incorporating automated disease detec tion systems into agricultural practices, we can significantly improve disease management strategies, reduce crop losses, and enhance food security. The findings of this study con tribute to the growing body of research on plant disease detection and provide valuable insights for further advancements in this field.