Proactive Peach Pest Control: Image Analysis and Real-Time Environmental Method

This study presents an intelligent peach pest prediction and control system that fuses deep‑learning image diagnostics with real-time IoT agro‑climatic sensing. A CNN trained on a large, expert-labeled dataset automatically detects key pests and diseases—brown rot, bacterial spot, aphids, and peach moth—achieving 92% classification accuracy. Concurrently, multi-point sensors stream temperature, humidity, soil‑moisture, and sunlight data to an LSTM forecasting model that learns environment-driven outbreak patterns. The two outputs are merged through a rule-based data‑fusion algorithm that grades risk and triggers alerts. Field trials in Suncheon and Gwangyang orchards confirmed that the integrated approach increases early-detection rates by 10% over image-only baselines, issues a warning an average of three days before visible symptoms appear, and enables targeted interventions that reduce chemical usage and damage. Certification testing by the Korea Institute of Lighting Technology further validated key performance targets, including≥87% predictive accuracy (achieved at 94.2 %), image analysis within 20 seconds, and sensor data processing within 1 minute. The modular edge-to-cloud architecture runs on cost-effective hardware, supports real-time dashboards and mobile notifications, and is readily extensible to other crops through transfer learning. By combining computer vision, time‑series analytics, and IoT, the proposed system offers a practical, scalable template for proactive, data‑driven crop protection that advances sustainable, precision agriculture. Future work will extend deployment through drone imagery, lighter edge models, and explainable‑AI modules to widen crop coverage and strengthen farmer trust.