Impact of Electrode Position on Forearm Orientation Invariant Hand Gesture Recognition
Authors:
Md. Johirul Islam,
Umme Rumman,
Arifa Ferdousi,
Md. Sarwar Pervez,
Iffat Ara,
Shamim Ahmad,
Fahmida Haque,
Sawal Hamid,
Md. Ali,
Kh Shahriya Zaman,
Mamun Bin Ibne Reaz,
Mustafa Habib Chowdhury,
Md. Rezaul Islam
Abstract:
Objective: Variation of forearm orientation is one of the crucial factors that drastically degrades the forearm orientation invariant hand gesture recognition performance or the degree of freedom and limits the successful commercialization of myoelectric prosthetic hand or electromyogram (EMG) signal-based human-computer interfacing devices. This study investigates the impact of surface EMG electr…
▽ More
Objective: Variation of forearm orientation is one of the crucial factors that drastically degrades the forearm orientation invariant hand gesture recognition performance or the degree of freedom and limits the successful commercialization of myoelectric prosthetic hand or electromyogram (EMG) signal-based human-computer interfacing devices. This study investigates the impact of surface EMG electrode positions (elbow and forearm) on forearm orientation invariant hand gesture recognition. Methods: The study has been performed over 19 intact limbed subjects, considering 12 daily living hand gestures. The quality of the EMG signal is confirmed in terms of three indices. Then, the recognition performance is evaluated and validated by considering three training strategies, six feature extraction methods, and three classifiers. Results: The forearm electrode position provides comparable to or better EMG signal quality considering three indices. In this research, the forearm electrode position achieves up to 5.35% improved forearm orientation invariant hand gesture recognition performance compared to the elbow electrode position. The obtained performance is validated by considering six feature extraction methods, three classifiers, and real-time experiments. In addition, the forearm electrode position shows its robustness with the existence of recent works, considering recognition performance, investigated gestures, the number of channels, the dimensionality of feature space, and the number of subjects. Conclusion: The forearm electrode position can be the best choice for getting improved forearm orientation invariant hand gesture recognition performance. Significance: The performance of myoelectric prosthesis and human-computer interfacing devices can be improved with this optimized electrode position.
△ Less
Submitted 16 September, 2024;
originally announced October 2024.
FORS-EMG: A Novel sEMG Dataset for Hand Gesture Recognition Across Multiple Forearm Orientations
Authors:
Umme Rumman,
Arifa Ferdousi,
Bipin Saha,
Md. Sazzad Hossain,
Md. Johirul Islam,
Shamim Ahmad,
Mamun Bin Ibne Reaz,
Md. Rezaul Islam
Abstract:
Surface electromyography (sEMG) signals hold significant potential for gesture recognition and robust prosthetic hand development. However, sEMG signals are affected by various physiological and dynamic factors, including forearm orientation, electrode displacement, and limb position. Most existing sEMG datasets lack these dynamic considerations. This study introduces a novel multichannel sEMG dat…
▽ More
Surface electromyography (sEMG) signals hold significant potential for gesture recognition and robust prosthetic hand development. However, sEMG signals are affected by various physiological and dynamic factors, including forearm orientation, electrode displacement, and limb position. Most existing sEMG datasets lack these dynamic considerations. This study introduces a novel multichannel sEMG dataset to evaluate commonly used hand gestures across three distinct forearm orientations. The dataset was collected from nineteen able-bodied subjects performing twelve hand gestures in three forearm orientations--supination, rest, and pronation. Eight MFI EMG electrodes were strategically placed at the elbow and mid-forearm to record high-quality EMG signals. Signal quality was validated through Signal-to-Noise Ratio (SNR) and Signal-to-Motion artifact ratio (SMR) metrics. Hand gesture classification performance across forearm orientations was evaluated using machine learning classifiers, including LDA, SVM, and KNN, alongside five feature extraction methods: TDD, TSD, FTDD, AR-RMS, and SNTDF. Furthermore, deep learning models such as 1D CNN, RNN, LSTM, and hybrid architectures were employed for a comprehensive analysis. Notably, the LDA classifier achieved the highest F1 score of 88.58\% with the SNTDF feature set when trained on hand gesture data of resting and tested across gesture data of all orientations. The promising results from extensive analyses underscore the proposed dataset's potential as a benchmark for advancing gesture recognition technologies, clinical sEMG research, and human-computer interaction applications. The dataset is publicly available in MATLAB format. Dataset: \url{https://www.kaggle.com/datasets/ummerummanchaity/fors-emg-a-novel-semg-dataset}
△ Less
Submitted 26 November, 2024; v1 submitted 3 September, 2024;
originally announced September 2024.
