Classification of Strawberry Plant Diseases Using Deep Learning Architecture for Optimal Results

Abstract

Strawberry (Fragaria x ananassa) cultivation plays an important role in local economies, especially in agrotourism regions like Bandung, Indonesia. Traditional disease identification methods, which rely on manual visual inspection by experts, are time-consuming, inconsistent, and infeasible for large-scale applications due to their subjective nature. To address these challenges, this study investigates the use of machine learning and computer vision models to automate and improve the accuracy of strawberry disease classification. We conducted a comparative analysis of nine machine learning models: DenseNet121, InceptionResNetV2, InceptionV3, MobileNetV2, ResNet50V2, VGG16, VGG19, YOLOv8n, and YOLOv11n. The dataset used in this study consists of 877 labeled images representing healthy and infected strawberry plants, collected from online sources (Kaggle, Roboflow) and field images. The data was preprocessed and split into training (70%), validation (20%), and testing (10%) subsets. Among all models, YOLOv8n and YOLOv11n achieved the highest classification accuracy of 94%, while also demonstrating fast inference speeds and relatively small model sizes. These results highlight their potential suitability for real-time disease detection in agricultural settings. The outcomes of this study aim to support the development of accessible, automated tools for strawberry disease diagnosis, particularly benefiting local farmers in agrotourism-based regions like Bandung.

Plant phenotyping relevance

イチゴ植物の病害状態を画像から分類・推定する深層学習手法が研究の中心であり、植物病害表現型の取得・自動化に該当する。

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.