Revisiting Noise Resilience Strategies in Gesture Recognition: Short-Term Enhancement in Surface Electromyographic Signal Analysis

Weiyu Guo, Ziyue Qiao, Ying Sun, Hui Xiong

Keywords: surface electromyography, gesture recognition, signal processing

TL;DR: STET is a new method for sEMG-based gesture recognition that enhances short-term features, improving accuracy, robustness to noise, and generalization across gestures, significantly outperforming existing approaches in noisy real-world environments.

Abstract: Gesture recognition based on surface electromyography (sEMG) has been gaining importance in many 3D Interactive Scene. However, sEMG is easily influenced by various forms of noise in real-world environments, leading to challenges in providing long-term stable interactions through sEMG. Existing methods usually struggle to improve generalizability or prediction reliability in the real scene, such as distinguishing similar gestures when facing various noises. To this end, in this paper, we propose a new method, called Short Term Enhanced Transformer (STET), which improves the precision and robustness against various common noisy scenarios by exploiting enhanced short-term features in time series. Compared with existing methods, STET possesses several unique merits: (1) preciseness, achieving high accuracy in different types of gestures; (2) robustness, mitigating the impact of noise in the real scene; and (3) generalization, being capable of doing gesture classification and hand joint angle regression. Finally, we have studied the performances of STET on the largest public sEMG data set including single-finger, multi-finger, wrist, and rest gestures. The results show that STET outperforms existing approaches by a large margin and can significantly improve robustness when facing various noises. More importantly, compared with best-competing approaches, the impact of noise on STET is reduced by more than 20\%. The extensive experiments also demonstrate that the short-term information is critical for sEMG-based gesture recognition and STET successfully exploits such information.

Primary Area: applications to neuroscience & cognitive science

Code Of Ethics: I acknowledge that I and all co-authors of this work have read and commit to adhering to the ICLR Code of Ethics.

Submission Guidelines: I certify that this submission complies with the submission instructions as described on https://iclr.cc/Conferences/2025/AuthorGuide.

Anonymous Url: I certify that there is no URL (e.g., github page) that could be used to find authors’ identity.

No Acknowledgement Section: I certify that there is no acknowledgement section in this submission for double blind review.

Submission Number: 10974