Artificial Intelligence-Based Hydroponic Plant Disease Detection System (Lactuca sativa)

Abstrak

Hydroponic cultivation of lettuce (Lactuca sativa) offers high water efficiency, yet productivity is frequently compromised by rapid disease spread and nutrient imbalances. Traditional manual monitoring is labor-intensive, time-consuming, and prone to subjective diagnostic errors, often leading to delayed interventions. This study aims to develop an automated, real-time disease detection system by integrating Deep Learning algorithms with an Internet of Things (IoT) architecture. The proposed method utilizes an optimized One-Stage Object Detector based on the YOLO framework, specifically designed for efficient deployment on edge computing devices. The model was trained and validated on a diverse dataset encompassing healthy plants, tip-burn, leaf spot, and nutrient deficiencies, employing rigorous data augmentation to ensure robustness against indoor lighting variability. Experimental results demonstrate that the system achieves a Mean Average Precision (mAP@0.5) of 94.8%, significantly outperforming conventional Support Vector Machine (SVM) approaches and standard detectors. The model maintains high detection accuracy even under complex background conditions. In conclusion, this research provides a viable, low-latency solution for precision agriculture, enabling growers to automate plant health monitoring and effectively minimize crop losses.