Precision farming: Using an IoT multimodal data-driven deep network to optimize irrigation in wheat crops

Abstract

Monitoring water demand and irrigation frequency in wheat crop can be challenging as it requires a deep understanding of the crop growth stage, environmental conditions, and soil moisture levels. However, with the advancements in the Internet of Things (IoT) and deep learning, it has become feasible to develop a data-driven approach capable of delivering highly accurate predictions. This research explores a potentially intelligent solution for tracking the frequency of wheat irrigation and its water requirements. The implemented setup integrates deep networks, such as convolutional neural network (CNN) and deep neural network (DNN), along with pre-trained networks like VGG16, VGG19, ResNet50, ResNet101, and MobileNet. The experimental data was collected through IoT-based sensors, including a digital camera, wind speed, soil moisture, air temperature, and relative humidity. During the process of gathering plant images, environmental factors (EF) were also recorded. The analysis outcomes indicated that the fusion of VGG16–EF features with CNN boosted the precision of the expected irrigation frequency and plant water status (96.2% for validation). These characteristics significantly outperformed those of other transfer learning features. Moreover, the hybrid model consisting of CNN_VGG19, CNN_EF, and DNN_EF attained the highest validation performance (97.9%), with precision, F-measure, recall, and intersection over union values of 98%, 97.9%, 97.9%, 95.9%, respectively. The planned framework outlines a roadmap for the automated detection of irrigation frequency and water status throughout a plant’s life cycle. In the future, the proposed methodology could play a crucial role in analyzing crop growth traits for precision farming and agricultural irrigation management.