What type of study is this?

September 10, 2025Open Access

Multimodal Feature Inputs Enable Improved Automated Textile Identification

Key Points

Textile classification accuracy improved significantly using multimodal feature inputs, enhancing precision and recall.
The ResNet50_V2 model achieved a precision of 0.929 and F1 score of 0.913, demonstrating effective classification.
Utilizing advanced feature extraction techniques and late fusion with convolutional neural networks optimizes results.
This methodology has important implications for diverse industrial applications, including fashion and quality control.

Abstract

This study presents an advanced framework for fabric texture classification by leveraging macro- and micro-texture extraction techniques integrated with deep learning architectures. Co-occurrence histograms, local binary patterns (LBPs), and albedo-dependent feature maps were employed to comprehensively capture the surface properties of fabrics. A late fusion approach was applied using four state-of-the-art convolutional neural networks (CNNs): InceptionV3, ResNet50V2, DenseNet, and VGG-19. Excellent results were obtained, with the ResNet50V2 achieving a precision of 0. 929, recall of 0. 914, and F1 score of 0. 913. Notably, the integration of multimodal inputs allowed the models to effectively distinguish challenging fabric types, such as cotton–polyester and satin–silk pairs, which exhibit overlapping texture characteristics. This research not only enhances the accuracy of textile classification but also provides a robust methodology for material analysis, with significant implications for industrial applications in fashion, quality control, and robotics.

Read Full Paperexternally

Perguntar à IA

Bookmark

View Full Paper