CupMoNN-SwBA: Temporal Citrus Disease Detection Using Coupled Modular Neural Network with Swarm Bipolar Algorithm

Citrus plants are highly valued worldwide for their fruits, including oranges, lemons, tangerines, and grapefruits, but their production is severely affected by diseases such as black spot, canker, and huanglongbing (greening). Accurate and timely detection of these diseases is critical to prevent yield loss and ensure fruit quality. Traditional disease detection methods are often time-consuming, subjective, and unable to track disease progression over time. To address these challenges, this research presents a novel framework, CupMoNN-SwBA: Temporal Citrus Disease Detection Using Coupled Modular Neural Network with Swarm Bipolar Algorithm. Initially, images are collected from the Citrus Leaves Prepared dataset, encompassing both healthy and diseased leaves. These images are fed into the preprocessing stage, where BitonicX Filtering (BiXF) is applied to reduce noise, enhance bright and dark disease patterns, and preserve fine leaf structures for improved visibility. Subsequently, the Adaptive Transformer Model (AdapTM) extracts adaptive local and global contextual features, capturing discriminative patterns necessary for accurate classification. The extracted features are then input to the Coupled Modular Neural Network (CupMoNN), which integrates common, localized, and disease-specific features from sequential images, effectively modeling temporal disease progression and improving robustness. Lastly, the Swarm Bipolar Algorithm (SwBA) uses the network weight parameter to balance the exploration of the global and exploitation of the local to maximize classification performance. The results of the experiment indicate that the proposed framework can serve well to detect and categorize the diseases in citrus and monitor the development of the symptoms over time. The CupMoNN-SwBA method is more accurate, robust, and sensitive to time than the existing approaches and provides a consistent instrument for monitoring and controlling of citrus diseases in real-time.