<p>The integration of evolutionary optimization techniques with machine learning models offers effective solutions for complex classification tasks. In this study, an optimized machine learning approach is proposed for the recognition of artistic materials in images. Specifically, the performance of the support vector machine algorithm is enhanced by tuning its key parameters using the Mountain Gazelle Optimizer, a recently developed bio-inspired metaheuristic algorithm. The proposed method captures fine-grained visual differences across various artistic media, such as gouache, oil, watercolor, acrylic, and charcoal. These media often present overlapping textures and tonal similarities that challenge conventional classification models. Experimental evaluations on both a publicly available image dataset and a real-world dataset composed of student-created paintings indicate that the proposed approach achieved 84% accuracy on testing data. The results suggest that evolutionary strategies can support improved model performance in fine-grained visual recognition tasks. Additionally, the method may assist educational technologies by helping students identify artistic materials and engage with visual content in structured learning contexts.</p>