Emerging Wearable Interfaces and Algorithms for Hand Gesture Recognition: A Survey

Jiang, Shuo; Kang, Peiqi; Song, Xinyu; Lo, Benny; Shull, Peter B.

doi:10.1109/rbme.2021.3078190

Cited by 96 publications

(37 citation statements)

References 220 publications

(203 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…[68] More information can be found in other references. [38,69,70] 2.1.2. Hierarchical Cluster Analysis Hierarchical cluster analysis (HCA) is an unsupervised algorithm of multivariate analysis for clustering, including two types: agglomerative (bottom-up approach) and divisive (top-down approach).…”

Section: Principal Component Analysismentioning

confidence: 99%

Augmenting Sensor Performance with Machine Learning Towards Smart Wearable Sensing Electronic Systems

Zhang

Suresh

Yang

et al. 2022

Advanced Intelligent Systems

View full text Add to dashboard Cite

Wearable sensing electronic systems (WSES) are becoming a fundamental platform to construct smart and intelligent networks for broad applications. Various physiological data are readily collected by the WSES, including biochemical, biopotential, and biophysical signals from human bodies. However, understanding these sensing data, such as feature extractions, recognitions, and classifications, is largely restrained because of the insufficient capacity when using conventional data processing techniques. Recent advances in sensing performance and system‐level operation quality of the WSES are expedited with the assistance of machine learning (ML) algorithms. Here, the state‐of‐the‐art of the ML‐assisted WSES is summarized with emphasis on how the accurate perceptions on physiological signals under different algorithms paradigm augment the performance of the WSES for diverse applications. Concretely, ML algorithms that are frequently implemented in the WSES studies are first synopsized. Then broad applications of ML‐assisted WSES with strengthened functions are discussed in the following sections, including intelligent physiological signals monitoring, disease diagnosis, on‐demand treatments, assistive devices, human–machine interface, and multiple sensations‐based virtual and augmented reality. Finally, challenges confronted for the ML‐assisted WSES are addressed.

show abstract

Section: Principal Component Analysismentioning

confidence: 99%

Augmenting Sensor Performance with Machine Learning Towards Smart Wearable Sensing Electronic Systems

Zhang

Suresh

Yang

et al. 2022

Advanced Intelligent Systems

View full text Add to dashboard Cite

show abstract

“…Furthermore, we considered the feature values of sEMG signals as input channels. In previous studies, we discovered that the feature values can be used to improve estimation accuracy during the training process [11,12]. A common problem that occurs when using fewer sensor inputs is insufficient gait information during the gait cycle, which affects the estimation accuracy [7].…”

Section: Normal Gait Cyclementioning

confidence: 99%

“…Recent studies have shown that sEMG is often used during rehabilitation because of its noninvasiveness. It does not involve any pain and discomfort and can be easily applied to the skin [10][11][12]. However, each person has different sEMG patterns.…”

Section: Introductionmentioning

confidence: 99%

Comparison of Deep Neural Network Models and Effectiveness of EMG Signal Feature Value for Estimating Dorsiflexion

et al. 2021

View full text Add to dashboard Cite

Robotic ankle–foot orthoses (AFO) are often used for gait rehabilitation. Our research focuses on the design and development of a robotic AFO with minimum number of sensor inputs. However, this leads to degradation of gait estimation accuracy. To prevent degradation of accuracy, we compared a few neural network models in order to determine the best network when only two input channels are being used. Further, the EMG signal feature value of average rate of change was used as input. LSTM showed the highest accuracy. However, MLP with a small number of hidden layers showed results similar to LSTM. Moreover, the accuracy for all models, with the exception of LSTM for one subject (SD), increased with the addition of feature value (average rate of change) as input. In conclusion, time-series networks work best with a small number of sensor inputs. However, depending on the optimizer being used, even a simple network can outrun a deep learning network. Furthermore, our results show that applying EMG signal feature value as an input tends to increase the estimation accuracy of the network.

show abstract

“…During the last decade, hand gesture recognition has been increasingly investigated by both academia and industry, following the versatility and easiness of its implementation. Indeed, thanks to technology progresses in the electronics fields, the recognition of hand movements can be now performed directly with body sensors [1], without the needing of a complex equipment, making it one of the best candidates for Human-Machine Interface (HMI) systems.…”

Section: Introductionmentioning

confidence: 99%

Hand Gestures Recognition for Human-Machine Interfaces: A Low-Power Bio-Inspired Armband

Mongardi

Rossi

Prestia

et al. 2022

IEEE Trans. Biomed. Circuits Syst.

View full text Add to dashboard Cite

Hand gesture recognition has recently increased its popularity as Human-Machine Interface (HMI) in the biomedical field. Indeed, it can be performed involving many different non-invasive techniques, e.g., surface Elec-troMyoGraphy (sEMG) or PhotoPlethysmoGraphy (PPG). In the last few years, the interest demonstrated by both academia and industry brought to a continuous spawning of commercial and custom wearable devices, which tried to address different challenges in many application fields, from tele-rehabilitation to sign language recognition. In this work, we propose a novel 7-channel sEMG armband, which can be employed as HMI for both serious gaming control and rehabilitation support. In particular, we designed the prototype focusing on the capability of our device to compute the Average Threshold Crossing (ATC) parameter, which is evaluated by counting how many times the sEMG signal crosses a threshold during a fixed time duration (i.e., 130 ms), directly on the wearable device. Exploiting the event-driven characteristic of the ATC, our armband is able to accomplish the on-board prediction of common hand gestures requiring less power w.r.t. state of the art devices. At the end of an acquisition campaign that involved the participation of 26 people, we obtained an average classifier accuracy of 91.9 % when aiming to recognize in real time 8 active hand gestures plus the idle state. Furthermore, with 2.92 mA of current absorption during active functioning and 1.34 ms prediction latency, this prototype confirmed our expectations and can be an appealing solution for longterm (up to 60 h) medical and consumer applications.

show abstract

Emerging Wearable Interfaces and Algorithms for Hand Gesture Recognition: A Survey

Cited by 96 publications

References 220 publications

Augmenting Sensor Performance with Machine Learning Towards Smart Wearable Sensing Electronic Systems

Augmenting Sensor Performance with Machine Learning Towards Smart Wearable Sensing Electronic Systems

Comparison of Deep Neural Network Models and Effectiveness of EMG Signal Feature Value for Estimating Dorsiflexion

Hand Gestures Recognition for Human-Machine Interfaces: A Low-Power Bio-Inspired Armband

Contact Info

Product

Resources

About