Flexible Strain Sensors With Multifusion Algorithm Models for Handwriting Recognition

Abstract

High sensitive flexible strain sensors were developed using styrene-ethylene–butylene-styrene (SEBS) G1657 and superconducting carbon black materials, which were placed on three fingers to capture handwriting through resistance changes. The designed sensor can withstand up to 600% strain, and the gauge factor (GF) can reach 19235.7, indicating extremely high responsiveness. For improved handwriting style recognition, a lightweight, modified Swin Transformer model was specifically designed for efficient classification. Experimental results demonstrated high classification accuracies of 99.73%, 99.18%, and 99.34% for digits, English letters, and Chinese characters, respectively, underscoring the model’s robustness and accuracy. These results represent a significant advancement in practical handwriting recognition, providing rapid and precise identification capabilities. Future efforts will focus on optimizing real-time detection algorithms, expanding recognition applications, and further enhancing the integration of conductive materials with machine learning techniques to achieve even greater accuracy and efficiency.