ABSTRACT Weather monitoring in agriculture is complex, as it requires predicting future atmospheric states that directly affect farming activities. In smart cities, farmers rely on accurate, up‐to‐date weather information to make informed decisions. Increasing climate variability has made temperature prediction more challenging than ever. Deep learning has recently emerged as a powerful approach for weather forecasting due to its superior performance over conventional methods and its ability to extract and classify features within a single architecture. This study explores the use of deep Convolutional Neural Network (CNN) models, specifically Visual Geometry Group 16 (VGG16) and MobileNet, with transfer learning for intelligent weather monitoring. The combination of MobileNet and VGG16 leverages transfer learning for accuracy‐driven and efficient operations. MobileNet is optimised for mobile and edge devices by delivering high‐performance results with lower computational and energy costs, while the deep architecture of VGG16 effectively identifies complex visual features across multiple tasks. Trained on a dataset of weather images, the proposed models accurately detect and classify different weather conditions, enabling farmers to make improved field decisions. Experimental results demonstrate strong performance, with VGG16 achieving 96.95% accuracy and MobileNet achieving 96.19% accuracy.
Tariq et al. (Thu,) studied this question.