Application of Deep Learning Based Gesture Recognition in the Functional Modification of Textiles

Xuechun Wang, Jifeng Qin

Article
2025 / Volume 8 / Pages 940-960
Received 3 July 2025; Accepted 9 September 2025; Published 11 December 2025
https://doi.org/10.31881/TLR.2025.940

Abstract
To address the inconvenience of interaction in smart textiles, we propose a Gesture-Interactive Electrochromic Textile System (GI-ECTS) that seamlessly combines high-performance PEDOT:PSS-based electrochromic yarns with a real-time gesture-recognition module powered by the lightweight deep learning architecture MobileNetV3. Experiments show that the embedded gesture-recognition model achieves an accuracy exceeding 95%, indicating the model’s potential for this application, while the functional yarns exhibit subsecond switching speeds and outstanding cycling stability. The results confirm the feasibility of a complete linkage from functional materials to intelligent algorithms, providing a viable technological pathway for next-generation interactive textiles.

Keywords
smart textiles, electrochromism, gesture recognition, deep learning