Leaf Prediction Models Using Deep Learning-based Bio-Inspired Algorithms

Precise forecasting of leaf characteristics is crucial for agricultural and environmental surveillance, supporting prompt disease identification and enhanced crop supervision. In order to improve prediction performance, this study offers a deep learning model optimized using three bio-inspired algorithms: Particle Swarm Optimization (PSO), Ant Colony Optimization (ACO), and Genetic Algorithms (GA). A dataset of 50,000 annotated leaf photos is used by the model, and preprocessing operations including scaling, normalization, and augmentation are performed. Significant gains were seen in the accuracy, precision, recall, and F1-score, which all reached 93.1%, 0.92, and 0.93, respectively. Furthermore, convergence rates were increased by hyper parameter tuning; GA achieved 95%, PSO 93%, and ACO 96%. The increases in performance were matched by a small increase in training hours. This model fared better than others with accuracies of 85.0%, 87.5%, and 89.8% when compared to earlier methods. These developments point to useful applications, such as more precise and effective real-time leaf health monitoring, which might boost crop yields and improve agricultural resource management, eventually promoting environmental sustainability.