Optimizing Deep Learning Models with Genetic Algorithms for Improved Classification and Identification of Wheat Leaf Diseases

Wheat is recognized as one of the most essential crops and a major food source worldwide. Ensuring an adequate balance of nutrient elements, such as Nitrogen (N), is crucial for healthy crop growth. The primary aim of this study is to classify nutrient deficiencies in leaf images. To achieve this, a genetic algorithm is applied to optimize the hyperparameters of a convolutional neural network for an image classification task. Convolutional neural networks have become increasingly prevalent in practical computer vision applications in recent years. The proposed convolutional network is trained with a wheat leaf images dataset. These images are categorized based on leaf color charts as 1, 2, 3, 4, 5, and 6 level. The primary goal is to enhance model accuracy by wisely searching through a predefined space of model architecture and training parameters. The initial accuracy of the model was 93.37% before applying the genetic algorithm for hyperparameter optimization. After optimization, the model achieved a final accuracy of 96.19%. The final model, trained using the best hyperparameters identified by the genetic algorithm, demonstrated improved performance on the validation set, indicating that hyperparameter optimization can significantly enhance CNN-based nutrient deficiency classification in wheat leaves. Earlier studies relied on a 4-strip LCC and achieved 94.22% accuracy, while the proposed method achieved a GA-optimized CNN with 6-strip LCC framework, improving accuracy to 96.19% and enabling finer nitrogen-level differentiation.