Ai-Enabled Embedded System for Plant Disease Detection and Precision Pesticide

Abstract

Agricultural productivity is heavily influenced by the timely detection of plant disease and efficient pesticide application. Traditional methods rely on manual inspection and broadspectrum spraying, which are labour-intensive and environmentally unsustainable. The existing system often lacks real-time responsiveness, precision targeting and integration between AI and embedded hardware. To overcome these limitations, we proposed an AI-enabled embedded system that detects plant diseases and applies pesticides with high accuracy and minimal waste. Our system uses a RetinexNet for illumination enhancement and DnCNN for noise reduction, implemented in Python. The disease classification results are transmitted via serial communication to an ESP32 controller programmed in Embedded C. The controller activates a relay-driven pesticide pump, displaying disease type on a 16×2 12C LCD and triggers an alert sound. The power is supplied via a regulated 12V battery system, ensuring field deployability. The experimental results show improved detection accuracy and a significant reduction in pesticide usage, demonstrating the system's potential for scalable eco eco-friendly smart farming.