: A stable agricultural economy, food security, and quality of life largely depend on efficient plant Pest Detection (PD). Rapid insect and pest infestations severely affect annual crop yields, and existing detection methods often face security and performance limitations. Globally, researchers are exploring scientific approaches that leverage Cloud Computing (CC) systems for pest detection. To address these challenges, several automation and Image Processing (IP)-based strategies have emerged in recent years. This study proposes a Hybrid Deep Learning (HDL) and Improved Feature Weighted Fuzzy Clustering (IFWFC) model for effective Pest Recognition (PR). Initially, preprocessing is performed using Z-Score Normalization (ZSN). The IFWFC process is then introduced, and data security is enhanced through a Henon Chaotic Map Encryption (HCME) technique. For feature selection, the Modified Cuckoo Search Algorithm (MCSA) is employed. Finally, pest recognition is achieved using a hybrid framework that combines HDL with Bidirectional Long Short-Term Memory (Bi-LSTM). In addition, a Faster Region Convolutional Neural Network (FRCNN) is integrated with Bi-LSTM for hybrid classification. To improve the performance further, the Improved Butterfly Optimization Algorithm (IBOA) is applied. Experimental evaluation using the IP102 dataset demonstrates that the proposed model achieves superior accuracy (ACC) compared to conventional Machine Learning (ML) methods, highlighting its effectiveness for automated pest recognition.
