AI-Based Early Disease Detection in Peach Crops: A Computer Vision Approach for Monilinia spp. and Taphrina deformans

Abstract

Abstract This study introduces a robust early disease detection system for peach crops, leveraging computer vision and artificial intelligence to address significant economic losses caused by Brown Rot (Monilinia spp.) and Leaf Curl (Taphrina deformans). The methodology comprises a structured approach, starting with the collection of a high-quality dataset of 640 images captured under real-world field conditions. These images, representing healthy and diseased fruits and leaves, underwent a rigorous preprocessing pipeline that included background removal, color space conversion, resizing, and contour detection to optimize them for model training. A Convolutional Neural Network (CNN) was developed and validated using k-fold cross-validation, achieving an outstanding accuracy of 90.28\% for fruit disease detection and 96.43\% for leaf disease detection during the validation phase. The model's final performance, evaluated with a confusion matrix, demonstrated a remarkable 100\% precision for Brown Rot in fruits and 96.4\% precision for Leaf Curl in leaves. These results confirm the system's reliability and its potential for practical application in precision agriculture. The project culminates in a functional web application, showcasing the viability of deploying deep learning solutions as accessible tools for farmers to facilitate timely and proactive crop management.

Plant phenotyping relevance

桃の果実・葉の病徴を画像から検出・分類するコンピュータビジョン手法を開発し、データセット、前処理、CNN、交差検証、性能評価まで中心的に扱っているため、植物病害フェノタイピング手法に該当する。

Code and data availability

The paper's 640-image peach disease dataset and CNN source code are not publicly deposited; both are stated to be available only upon reasonable request from the corresponding author. No public URL or repository for the dataset, images, code, or trained model is provided in the supplied blocks.

No evidence-backed public reproduction asset is currently recorded.