Edge AI and CNN for Real-Time Plant Disease Detection using IOT

The integration of Edge Artificial Intelligence (Edge AI) with Convolutional Neural Networks (CNNs) presents a transformative approach to real-time plant disease detection in agriculture. This paper proposes a lightweight deep learning model tailored for deployment on low-power edge devices such as Raspberry Pi and NVIDIA Jetson Nano. Utilizing the MobileNetV2 architecture, the model processes leaf images locally, enabling immediate diagnosis without relying on cloud infrastructure. A hybrid dataset, including the PlantVillage benchmark and a custom collection of sweet lemon leaf images, was used to train and evaluate the system. Results demonstrate an accuracy of 98.6%, with inference times of 80ms and 120ms on Jetson Nano and Raspberry Pi respectively, showcasing the feasibility of realtime operation. This approach enhances data privacy, reduces latency, and minimizes bandwidth consumption—critical factors for remote agricultural regions. The paper also compares the proposed model with traditional CNN architectures and cloud-based implementations, highlighting the advantages of Edge AI in smart farming. Future enhancements such as federated learning, drone integration, and explainable AI are also discussed.