Agriculture plays a vital role in economic growth, and the rapid advancement of Artificial Intelligence (AI) has enabled increasingly automated and efficient agricultural practices. In this study, we focus on the classification of tomato leaf diseases using image data from 10 classes. In addition, we employed a Conditional Generative Adversarial Network (cGAN) to generate synthetic images for data augmentation, achieving an optimal average FID score of 11.21, which indicates high-quality generated images. We then proposed a hybrid classification framework that extracts deep features from a custom-designed network, DCNN-GAPNet, and uses them as input to traditional machine learning models such as SVM and kNN. Experimental results demonstrate that this hybrid approach outperforms deep CNNs alone. Notably, the DCNN-GAPNet + SVM model trained on the cGAN-augmented dataset achieved the highest classification performance, significantly surpassing models trained on the original dataset.