Development Coding to Support a Smart Greenhouse System for Household Vegetable Gardening with Convolutional Neural Network

Abstract

This study presents the outcomes of the “Coding Coaching and Coding Learning” project, which aimed to enhance technological literacy and programming proficiency among secondary school students and teachers. A smart greenhouse system prototype was developed to facilitate household vegetable cultivation, integrating Internet of Things (IoT) technologies through a project-based learning (PBL) approach. Participants received training in microcontroller programming, sensor integration, and automation system design using the Arduino ESP32 platform and the Blynk application for remote monitoring and control. The prototype enabled real -time environmental management of temperature, humidity, light, and soil moisture. A 35-day cultivation trial involving lettuce, spring onions, and holy basil demonstrated significant improvements in IoT and coding knowledge (scores increased from 2.1 to 4.3, p < 0.05). The greenhouse maintained optimal conditions, produced an average of 750 grams per cycle, and saved approximately 30% water compared to traditional irrigation. User satisfaction scores averaged 4.8/5. Additionally, an AI-based CNN model detected leaf abnormalities achieved an accuracy of 87.5% on 500 images, highlighting the potential for small-scale smart farming systems in education and practical application.