In recent decades, the area under maize cultivation has increased due to its essential role in the food cycle for humans, livestock, and poultry. However, plant diseases pose a threat to food safety and can significantly reduce both the quality and quantity of agricultural products. Accurate and timely diagnosis of such diseases presents many challenges. This research proposes a novel approach based on a deep neural network to address these challenges. Due to the limited amount of available data, the transfer learning technique is employed using two well-known architectures. A new effective model is developed by combining pre-trained MobileNetV2 and Inception networks, chosen for their strong performance in object detection tasks. The convolutional layers of MobileNetV2 and the Inception modules are arranged in parallel in the early layers to extract crucial features. The model is trained and evaluated on the maize leaf subset of the publicly available PlantVillage dataset, which contains images of diseased and healthy plant leaves. Additionally, the class imbalance problem is addressed through a data augmentation strategy. The proposed method demonstrates superior performance compared to other state-of-the-art models published in recent years, achieving an accuracy of approximately 96.10 %. In summary, the experimental results validate the effectiveness and robustness of the proposed method in diagnosing plant leaf diseases

Keywords