This chapter presents a new hybrid deep learning (DL) model for the classification of apple leaf diseases. The new proposed model is called DenseNet-CEGJO. The model integrates DenseNet neural netwok with the Copula Entropy-based Golden Jackal Optimization (CEGJO) algorithm to optimize the feature selection (FS) process and enhance the performance of the proposed model. Apples are fundamental to global agriculture, contributing significantly to food security, nutrition, and sustainable development. So, pre-detecting the apple leaf diseases become a very important issue in the agriculture sector. Traditional models of classifying apple leaf diseases rely on manual inspection, which is often time-consuming, subjective, and error-prone. In contrast, advancements in Artificial Intelligence (AI), particularly Machine Learning (ML) and DL, enable automated, accurate, and timely disease detection, improving the classification accuracy. In this chapter, the framework of the methodology of the proposed DenseNet-CEGJO includes five main phases, starting from from the dataset preparation to the final classification of various apple leaf diseases such as rust, scab, and multiple diseases. By optimizing the FS process and minimizing the number of parameters, the proposed DenseNet-CEGJO model achieves a 99.9% classification accuracy, overcoming existing real-time detection challenges. Its high performance across metrics, including inference time, accuracy, precision, recall, FLOPs, and F1-score, makes it suitable for mobile and portable devices use. This efficient solution supports real-time disease identification on smartphones and portable devices in agricultural settings, contributing to increased apple yield and the promotion of sustainable farming practices.