Reducing the latency between disease detection and remedial action is critical for preventing crop loss in precision agriculture. This paper introduces the Smart Automated Garden Entity, an AIoT system developed for real-time monitoring and autonomous management of plant health. While traditional classifiers often rely on static image processing, Smart Automated Garden Entity utilizes a Convolutional Neural Network (CNN) optimized for edge deployment, enabling live video inference across 38 distinct plant pathologies. The system employs a multi-threaded architecture to synchronize visual detection with environmental data—including soil moisture, temperature, and humidity—collected at 500 millisecond intervals via an Arduino-based sensor node. By cross-referencing visual symptoms with environmental context, the system improves diagnostic accuracy and reduces false positives common in variable lighting. When stress vectors are identified, Smart Automated Garden Entity executes autonomous interventions, such as triggering irrigation relays, while simultaneously using the Gemini API to provide localized remedial advice to a user dashboard. Experimental results demonstrate a response time of under 1.5 seconds from detection to intervention, suggesting that the system is a viable low-latency solution for automated small-scale agriculture.
HARSHADITH et al. (Sun,) studied this question.