This study aims to present an integrated fabric classification system aimed at enhancing personalized, season-aware fashion recommendations. The objective is to accurately identify suitable fabric types – such as cotton for summer and wool for winter – using lightweight machine learning techniques, thereby enabling intelligent outfit suggestions based on fabric properties and climatic relevance.
The proposed framework uses a hybrid feature extraction strategy that combines RGB and luminance-based statistical descriptors with texture features derived from the Gray Level Co-occurrence Matrix (GLCM). Random Forest (RF)-based feature selection is used to select key features and reduce dimensionality. The optimized features are then classified using support vector machine (SVM) and K-Nearest Neighbors (KNN). The data set comprises iBUG images augmented with custom samples to ensure balanced representation of major fabric types – Velvet, Silk, Cotton and Denim.
Experimental evaluation reveals that the SVM classifier with a polynomial kernel achieves an accuracy of 97.8%, demonstrating the robustness of the proposed model. The KNN classifier, using the Chebyshev distance, strikes a favorable balance between precision and computational efficiency, making it suitable for real-time or resource-constrained applications.
The system offers a scalable solution for fashion retail, e-commerce and smart textiles. Linking fabric identification with seasonal preferences enhances recommendation accuracy, customer satisfaction and inventory management.
The novelty lies in the unified integration of color, texture and statistical features with machine learning methods, offering a precise and computationally efficient approach to fabric recognition.
