Implementation of YOLO11 for Disease Detection in Strawberry Plants Based on Android Application

Abstract

A Diseases in strawberry plants, particularly leaf spot and powdery mildew, represent major challenges that can diminish fruit quality and production quantity. Manual diagnosis by farmers is often subjective, time-consuming, and prone to error. This research aims to develop an automated strawberry disease detection system by implementing and comparing three variants of the latest deep learning architecture, YOLO11n (Nano), YOLO11s (Small), and YOLO11m (Medium), into an Android-based application. The results indicate that the YOLO11n (Nano) variant, as the baseline, provides the most optimal performance for mobile use, achieving a mean Average Precision (mAP@50) of 91.7%, a precision of 0.888, and a recall of 0.841. After integration into Android devices using the TensorFlow Lite format, the model recorded a real-time inference time ranging from 104-125 ms at a speed of 8 FPS. This study contributes an empirical framework for deploying cutting-edge deep learning models on resource-constrained edge devices, establishing that YOLO11n effectively bridges the gap between state-of-the-art detection accuracy and mobile operational efficiency. Furthermore, it provides a practical roadmap for the digital transformation of early-stage crop monitoring, enabling farmers to perform reliable, real-time diagnostics directly in the field.