Integrated E-Commerce Platform for Fresh Vegetables with YOLOv8-Based Lettuce Disease Detection

Abstract

This paper presented the design, implementation, and functional evaluation of an integrated digital platform for fresh-vegetable retail that combines electronic commerce (e-commerce) functionality with automated, image-based lettuce disease detection. The system was motivated by the dual challenge faced by small-scale vegetable vendors and hydroponic growers in Indonesia: the need for a low-cost digital sales channel and the need for early, low-cost diagnosis of common leaf diseases that reduce crop quality and yield. The platform was built as three cooperating components: a Go/Fiber RESTful backend with JSON Web Token (JWT) authentication and Midtrans payment-gateway integration, a Next.js administrative dashboard for catalogue and order management, and a React Native (Expo) customer mobile application that embeds a camera-based disease-detection feature. Disease detection was implemented by forwarding captured leaf images from the mobile client, through the backend, to a dedicated Python inference microservice running a YOLOv8 object-detection model, which returned the predicted disease class, confidence score, and treatment recommendation. WhatsApp-based order notifications were integrated through a webhook-driven messaging gateway. Black-box functional testing confirmed that all core e-commerce workflows and the detection pipeline operated as designed across the three repositories. The results indicated that a modular, microservice-based architecture is a practical and reproducible way to combine transactional e-commerce features with deep-learning-based plant health monitoring on resource-constrained mobile devices.