Fusion of MobileNet and GRU: Enhancing Remote Sensing Applications for Sustainable Agriculture and Food Security

Remote sensing technologies have become integral to modern agriculture, allowing for precise monitoring of crop health, soil moisture, and environmental factors. The fusion of spatial and temporal data through advanced machine learning models has paved the way for smarter agricultural decision-making. This paper presents a novel MobileNet-GRU fusion model designed for enhancing remote sensing applications in sustainable agriculture and food security. The model combines MobileNet’s lightweight convolutional neural network (CNN) architecture with gated recurrent unit (GRU) to process spatial and temporal data. The MobileNet component efficiently extracts spatial features from satellite imagery, such as crop health, soil moisture patterns, and environmental conditions, while the GRU captures temporal dependencies, such as seasonal changes and weather fluctuations. This fusion model has been tested across various crop types and environmental factors, achieving a high training accuracy of 93.0%, validation accuracy of 92.5%, and test accuracy of 91.5%. Moreover, the model demonstrates low mean absolute error (MAE) values for critical agricultural metrics, such as 0.053 for soil moisture and 0.019 for temperature prediction. The evaluation shows that this hybrid model outperforms other combinations of CNNs and RNNs in both performance and computational efficiency, making it a robust tool for real-time agricultural monitoring and decision-making. The fusion of MobileNet and GRU provides an integrated approach for precision farming by effectively analyzing large-scale, high-dimensional remote sensing data, improving yield predictions and resource management.