Detection of Paddy Leaf Disease Using Fine Element Integrated Neural Network with Pine Cone Optimization

Paddy (Oryza sativa) is a staple crop globally, providing essential nutrition as well as financial stability. However, a number of leaf diseases have a major impact on its yield, which can reduce yield and quality if not detected early. Early and precise identification of paddy leaf diseases is essential for timely intervention and crop protection. Traditional detection methods often rely on manual inspection or basic machine learning models, which suffer from limited accuracy, sensitivity to illumination changes, and poor generalization across diverse disease patterns. To overcome these limitations, this work presents Detection of Paddy Leaf Disease using Fine Element Integrated Neural Network with Pine Cone Optimization (FinEIN-PiCO). Initially, paddy leaf image data are collected from standard rice leaf disease datasets. The images undergo pre-processing using Memory Boosted Guidance Filtering (MBGFil) to enhance disease-specific features and normalize lighting conditions. Feature extraction is then performed using the Dual Selective Fusion Transformer Network (DSFusiTN), which captures both local and global disease patterns for discriminative representation. The extracted features are fed into the Fine Element Integrated Neural Network (FinEIN) to detect and classify paddy leaf diseases by modeling fine-grained feature interactions. To further improve performance, Pine Cone Optimization (PiCO) is employed to optimize network weights and the loss function, enhancing convergence, classification accuracy, and robustness. With a precision of 99.45%, recall of 99.38%, F1-score of 99.72%, sensitivity of 99.64%, and accuracy of 99.82%, the experimental results demonstrate that the suggested FinEIN-PiCO methodology outperforms other current methods for paddy leaf disease identification. This comprehensive approach offers a significant advancement for smart agriculture and disease management in paddy cultivation