Multi-Class Tuber Plant Leaf Disease Detection Using Hybrid Deep Learning Framework for Real Time Application

Abstract

The productivity and sustainability of agriculture depend on the early diagnosis of plant diseases, particularly for root crops such as potatoes, tomatoes, and carrots. The hybrid deep model proposed in this study employs a Convolutional Neural Network (CNN) architecture to provide precise and realtime multi-categorization of several leafy tuber crop diseases. A complete dataset of 20,657 labeled photos from 16 diseases was used to train the model, and regular classes were employed. Our goal is to build a scalable deep learning model that can diagnose multiple tuber crop diseases in real time, reduce the need for manual surveys through a cost-effective web tool, and support farmers with early detection for smarter and more sustainable crop management. The proposed CNN architecture, which uses convolutional, pooling, and fully connected layers that are modified by the Adam optimizer, was developed using TensorFlow and Keras. The constructed model was highly successful in detecting widespread illnesses such as early blight, late blight, and other tomato and carrot leaf diseases, as demonstrated by its 92.2% total accuracy rate. Being developed as a web application later on, the system gave farmers and agri-parties an efficient and economical diagnosis tool. Based on knowledge, early detection of disease, lower reliance on manual surveys, and crop management decisions, this work encourages precision agriculture.

Plant phenotyping relevance

植物葉の画像から病害状態を推定する深層学習手法の開発・評価が中心であり、植物フェノタイピング手法に該当する。

Code and data availability

公開本文の所在を確認できませんでした。非公開または購読が必要な可能性があります。

No evidence-backed public reproduction asset is currently recorded.