Article: Classification and trend analysis of creative packaging pattern design styles based on clustering algorithm Journal: International Journal of Information and Communication Technology (IJICT) 2026 Vol.27 No.84 pp.78 - 101 Abstract: To address the lack of systematic classification and trend analysis for creative packaging pattern design styles, this paper proposes a novel multimodal deep clustering framework. The method integrates convolutional neural network texture features, vision transformer global structure features, and simple linear iterative clustering balanced iterative reducing and clustering using hierarchies colour features, fusing them adaptively with learnable weights for dynamic feature fusion and end-to-end jointly optimised with deep embedding clustering to achieve unsupervised style classification. On the self-built packaging dataset containing 9,000 images, the method consistently achieves an accuracy of 0.768 and an F1-score of 0.755, thus significantly outperforming existing approaches (<i>p</i> < 0.05). The results reveal design evolution trends from 2015 to 2024, providing valuable data-driven support for design decision-making, trend prediction, and practical design practice. Inderscience Publishers - linking academia, business and industry through research

Title: Classification and trend analysis of creative packaging pattern design styles based on clustering algorithm

Authors: Xiaojie Jiao

Addresses: The School of Art, Zhengzhou University of Science and Technology, Zhengzhou, 450064, China

Abstract: To address the lack of systematic classification and trend analysis for creative packaging pattern design styles, this paper proposes a novel multimodal deep clustering framework. The method integrates convolutional neural network texture features, vision transformer global structure features, and simple linear iterative clustering balanced iterative reducing and clustering using hierarchies colour features, fusing them adaptively with learnable weights for dynamic feature fusion and end-to-end jointly optimised with deep embedding clustering to achieve unsupervised style classification. On the self-built packaging dataset containing 9,000 images, the method consistently achieves an accuracy of 0.768 and an F1-score of 0.755, thus significantly outperforming existing approaches (p < 0.05). The results reveal design evolution trends from 2015 to 2024, providing valuable data-driven support for design decision-making, trend prediction, and practical design practice.

Keywords: clustering algorithm; packaging design; style classification; trend analysis; multimodal features.

DOI: 10.1504/IJICT.2026.155478

International Journal of Information and Communication Technology, 2026 Vol.27 No.84, pp.78 - 101

Received: 30 Mar 2026
Accepted: 19 May 2026
Published online: 03 Aug 2026 *

Title: Classification and trend analysis of creative packaging pattern design styles based on clustering algorithm

Keep up-to-date