A Hybrid Edge-Cloud IoT Framework for Real- Time Plant Disease Detection using Multimodal Data Fusion

Abstract— The rising rate of plant diseases poses a significant
risk to global agriculture, leading to reduced yield and
financial loss. The current approaches of disease detection are
through manual visual inspection, or processing using cloud
images, with all methods struggling with latency, scalability
and power consumption limitations; especially in remote rural
areas. This work presents a new, light-weight, and effective
IoT-based system called SmartSense+, which is aimed for realtime
detection of plant diseases using hybrid edge-cloud
infrastructure and multimodal data fusion. The system
incorporates NodeMCU and ESP32-CAM modules for image
acquisition at the edge level and for monitoring environmental
parameters such as temperature, humidity, and soil moisture.
By using adaptive thresholding and light-weight feature
extraction processes on the edge device, only the necessary data
is sent through MQTT protocol to a cloud server. The cloud
layer uses a Decision Tree classifier with decision-level data
fusion augmented features, which are combined with sensor
data to enhance the prediction capability. Experimental results
validate that the devised system provides a detection accuracy
of 92.3% with considerably smaller data transmission size and
latency in comparison with traditional cloud-only and TinyML
techniques. In addition, the system facilitates real-time alerts
and includes an easy-to-use mobile and web dashboard for
farm surveillance. The contribution of this paper is a scalable,
energy-efficient, and cost-effective solution optimized for small
and medium-scale farms. The suggested hybrid IoT platform is
highly promising for solving actual agricultural problems by
facilitating early and precise detection of plant diseases to
protect plant health and develop more sustainable agriculture.
Keywords— Plant Disease Detection, Edge Computing, cloud
computing, NodeMCU, Internet-of-Things