Edge AI-Powered System Architecture for Aloe Vera Plant Disease Detection

Aloe Vera has been widely cultivated for medicinal and cosmetic purposes, but its productivity is affected by different leaf diseases. They need to be detected early and accurately to avoid crop loss and to keep plants healthy. Conventional disease identification techniques are based on manual inspection, which is both time-consuming and predisposed to errors. Therefore, to cope with this delinquency, this study presents an edge AI-based system architecture for real-time aloe vera plant disease detection, which leads to more efficient and accurate detection. An Aloe Vera disease dataset was employed to train a Convolutional Neural Network (CNN), which was then deployed on an edge device to perform real-time inference. Environmental monitoring using IoT sensors is also part of the architecture. Experimental results specify that the proposed system can perceive diseases with high accuracy while considerably reducing latency compared to cloud-based methods. The proposed Aloe Vera leaf disease classification model, based on ResNet50, achieved 99.15% accuracy, 99.20% precision, 99.21% recall, and a 99.20% F1 score, ensuring high classification performance. The deployment of the quantized TFLite model on Raspberry Pi 4 B enables real-time disease detection with an inference latency of 4,922 ms (~4.9s) and a reduced model size of 23.4MB (INT8), making it suitable for edge computing applications in precision agriculture. Fine-grained deep learning with Edge AI empowers Real-Time Decision Making in Resource-Constrained Environments. This study provides a solution for Aloe Vera disease detection, characterized by low latency, energy efficiency, and scalability, emphasizing a tool for smart agriculture applications.