Development of a model for collecting data on cantaloupe melons in greenhouses using IoT and AI

The direct and continuous monitoring of fruit growth in general, and cantaloupe melons in particular, within greenhouse environments is a critical requirement. Continuous observation enables regular assessment of the growth status of cantaloupe fruits. Consequently, the crop management system can access accurate information about the developmental condition of the fruits and propose specific solutions to promote growth according to the farmer’s expectations. This paper proposes a model for collecting real-time data directly from the field using an Internet of Things (IoT) system. The designed model includes mechanical structures and motors to control the camera’s movement. The camera is programmed to move to predefined positions 10 cm apart, within a total travel range of 3 m. Additionally, it can incrementally rotate by 10[Formula: see text] along the vertical axis from 0[Formula: see text] to 180[Formula: see text] to capture data from multiple viewing angles. The collected data are automatically transmitted and stored on a central server twice daily, at 6 a.m. and 4 p.m. In parallel, we apply deep learning-based object recognition algorithms to identify cantaloupe fruits in images. The resulting annotated dataset can subsequently be used for future applications to evaluate cantaloupe fruit growth. Experimental results show that 100% of the data was successfully transmitted and stored on the server. The fruit detection model achieved an overall recall of 87%. This study is novel in employing a movable IoT-camera system integrated with AI, offering greater flexibility and richer visual monitoring compared to traditional static sensor-based methods. In addition, the proposed system improves monitoring accuracy, while significantly reducing time and operational costs, providing an efficient and practical solution for real-time crop management.