Key points are not available for this paper at this time.
The demand for sustainable agricultural practices continues to rise, highlighting the need for precise crop and fertilizer recommendations to optimize yield while minimizing environmental impacts. Traditional methods often struggle to achieve this precision due to limitations in integrating diverse data sources and efficiently processing complex interactions. This paper addresses this challenge by introducing a novel methodology that seamlessly integrates multimodal data sources, including NPK values, moisture content, image analysis, and geographical information, to provide accurate and customized recommendations for crops and fertilizers. The proposed methodology employs a systematic fusion of Frequency patterns, entropy patterns, S Transform components, and convolutional components extracted from the data sources. These components are then transformed into Bidirectional Gated Recurrent Unit (BiGRU) features, which are meticulously selected using an exclusively designed Ant Lion Fuzzy Principal Component Analyzer (ALFPCA). The selected features are further processed through our Graph Convolutional FPMax Model (GCFPMax) for crop recommendations and our Recurrent FPMax Model (RFPMax) for fertilizer recommendations. The findings demonstrate that our proposed model significantly enhances the precision of crop recommendations by 3.5%, accuracy by 4.9%, recall by 2.5%, area under the curve (AUC) by 3.9%, and specificity by 2.9% while concurrently reducing the delay by 8.3%. Similarly, the precision of fertilizer recommendations improved by 1.9%, accuracy by 2.5%, recall by 3.5%, AUC by 3.9%, and Specificity by 4.5%, with a delay reduction of 8.5%. The significant improvements in recommendation accuracy and efficiency highlight our approach's transformative impact on achieving sustainable farming practices.
Bhattacharya et al. (Mon,) studied this question.