Plant diseases pose a serious threat to agricultural productivity and global food security, particularly for essential crops such as maize. Early and accurate identification of maize leaf diseases is therefore critical for reducing crop losses and improving farming efficiency. This study proposes an optimized deep learning framework for maize leaf disease classification by integrating Bayesian Optimization with CNN models. BO is employed to automatically determine optimal hyperparameter settings, including convolutional filter sizes, learning rate, and dense layer units, thereby enhancing the model’s predictive performance. Here, an ensemble strategy based on Bayesian-optimized CNN models (Ensemble_BO) is developed. The proposed framework achieved classification accuracies of 96.51% on training data and 92.77% on testing data which outperforms other models..