Accurate wheat species classification is critical for agricultural quality control, supply chain optimization, and breeding research. Manual identification is labor-intensive and error-prone, necessitating automated solutions. This study evaluates Support Vector Machines (SVM), VGG16, and a Hybrid Model that integrates handcrafted statistical features with deep learning representations for enhanced classification accuracy. Using a dataset of five wheat species, data augmentation techniques were applied to improve model generalization. Results show that the Hybrid Model outperforms SVM and VGG16, achieving 96.3% accuracy. These findings highlight the advantage of combining traditional feature extraction and deep learning for scalable and efficient automated grain classification. Future research will focus on dataset expansion for broader agricultural applicability.