This study proposes a novel hybrid model for dry bean variety classification that synergizes deep learning with pre-extracted features to bolster classification accuracy. Our approach integrates a compact one-dimensional Convolutional Neural Network (1DCNN) with 16 pre-extracted features, encompassing 12 geometric and 4 shape-based features. The 1DCNN excels at capturing spatial hierarchies and intricate patterns within the data, while the pre-extracted features offer valuable insights into the beans morphology and shape. Experimental results demonstrate that our hybrid model achieves a classification accuracy of 94.79%, surpassing traditional machine learning techniques like Support Vector Machine (93.13%) and deep learning models DBANN1 (93.07%) and DBANN2 (93.44%). The enhancement underscores the transformative potential of combining deep learning with carefully extracted morphological features. By leveraging both approaches, our hybrid model effectively addresses the challenges inherent in dry bean variety classification, providing a robust solution for sustainable and high-quality seed production.