This work seeks to contribute to the realization, by 2030, of the United Nations’ Sustainable Development Goal number two: “End hunger, achieve food security and improved nutrition and promote sustainable agriculture.” To contribute to the attainment of this goal, farm crops must be in a healthy state and maintained in healthy conditions. So, a round-the-clock monitoring of plant conditions is vital. There have been some plant monitoring techniques, such as the conventional monitoring technique, in which human is involved with the physical inspection of crops on farmlands. Satellite monitoring of crops has been used in some other instances but are expensive and limited to large scale farmlands. To make crop monitoring technology available to all including peasant farmers in developing nations, the adoption of convolution matrix of artificial neural network and Internet of Things (IoT) gives birth to this new approach. The developed system demonstrated high accuracy of 0.51 from the classification report obtained while detecting crop diseases and defects, collecting real-time environmental data. The macro average value of 0.56 was obtained for precision of the developed model for the 21 samples considered. Classification recall weighted average value of 0.51 and weighted f1-score of 0.49 were obtained simultaneously. For the model’s hardware, key components used by the system included an IoT-based sensor network, a camera system utilizing YOLOv8 for image processing, an automated response system, and a cloud-based platform for remote monitoring. Careful assembly of these components formed a formidable remote monitoring and reporting system that seeks to ease and improve methods of plant monitoring. For the analysis of the plant leaves, a neural network processes the image, and comparisons are made with the stored feature of healthy crop leaves. Confusion and classification results showed significant potential for enhancing crop management practices, improving resource utilization, and enabling data-driven agricultural decision-making. Validation of the AI-enhanced model was carried out using field data logged in cloud by means of IoT. All these operations are displayed on a liquid crystal display unit of the model. The successful implementation of this technology serves as a model for the broader adoption of smart farming techniques, with implications for improving agricultural productivity and sustainability. While further long-term testing is recommended, this work presents a significant contribution to the field of precision agriculture, offering promising solutions to critical food crisis challenges in the world.
Adewuyi et al. (Mon,) studied this question.