Hand Gesture Recognition based on Surface Electromyography using Convolutional Neural Network with Transfer Learning Method

Chen, Xiang; Li, Yu; Hu, Renfang; Zhang, Xu; Chen, Xun

doi:10.1109/jbhi.2020.3009383

Cited by 126 publications

(70 citation statements)

References 24 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…By using the four repetition gestures of new subject, the average accuracy of new subject recognition were 58.41% for the TL-based strategy and 54% for the non-TL based strategy, and our proposed method obtained 74.62% for the TL-based strategy and 59.47% for the non-TL based strategy. Chen [ 36 ] introduced an HD-sEMG based transfer learning method using a convolutional neural network. By pre-training based on a 30 multi-mode gesture dataset, the performed transfer learning based on one repetition gesture obtained more than 75% accuracy for both new subject and new gesture recognition, reached 90% accuracy when more than two repetitions were included and got an average accuracy of 91.18% for new gesture recognition in DB-a.…”

Section: Discussionmentioning

confidence: 99%

“…A further study by Cote-Allard [ 35 ] validated the improved PNN for the recognition of eleven new gestures, and the proposed method achieved a recognition accuracy of 49.41% compared to 46.06% without transfer learning. Chen [ 36 ] introduced an HD-sEMG based transfer learning method using a convolutional neural network. By pre-training based on a multi-mode dataset that contains 30 gestures, keeping some of the parameters consistent with the target network, and performing transfer learning based on two repetitions of training data for each gesture, the recognition accuracy of new subjects and 10 new gestures could reach 90%.…”

Section: Related Workmentioning

confidence: 99%

“…Although relevant studies on sEMG-based hand gesture recognition have verified that transfer learning can improve the generalization of the source network for new subjects or new gestures recognition to some degree, some issues, such as the heavy training burden [ 37 ], strict condition of use [ 36 ] and low generalization of the algorithm remain unsolved [ 19 , 35 , 38 ]. Yosinski [ 39 ] explored the transferability of features in the deep neural network, and the results proved that the transferability of features decreases when transferring a higher task-specific layer.…”

Section: Related Workmentioning

confidence: 99%

“…First, it is challenging to recognize hand gestures of a new subject due to subject-dependent characteristic of sEMG. Second, it is difficult to train a new classifier with a high accuracy when only a small amount of target training data available [ 35 , 36 , 38 ]. It is important to have a high recognition accuracy of the classifier while only using a small number of gesture repetitions of new subjects and new gestures for training.…”

Section: Instantaneous Hand Gesture Recognition Neural Network Andmentioning

confidence: 99%

See 3 more Smart Citations

Surface EMG-Based Instantaneous Hand Gesture Recognition Using Convolutional Neural Network with the Transfer Learning Method

Zhao

Wang

et al. 2021

Sensors

View full text Add to dashboard Cite

In recent years, surface electromyography (sEMG)-based human–computer interaction has been developed to improve the quality of life for people. Gesture recognition based on the instantaneous values of sEMG has the advantages of accurate prediction and low latency. However, the low generalization ability of the hand gesture recognition method limits its application to new subjects and new hand gestures, and brings a heavy training burden. For this reason, based on a convolutional neural network, a transfer learning (TL) strategy for instantaneous gesture recognition is proposed to improve the generalization performance of the target network. CapgMyo and NinaPro DB1 are used to evaluate the validity of our proposed strategy. Compared with the non-transfer learning (non-TL) strategy, our proposed strategy improves the average accuracy of new subject and new gesture recognition by 18.7% and 8.74%, respectively, when up to three repeated gestures are employed. The TL strategy reduces the training time by a factor of three. Experiments verify the transferability of spatial features and the validity of the proposed strategy in improving the recognition accuracy of new subjects and new gestures, and reducing the training burden. The proposed TL strategy provides an effective way of improving the generalization ability of the gesture recognition system.

show abstract

Section: Discussionmentioning

confidence: 99%

Section: Related Workmentioning

confidence: 99%

Section: Related Workmentioning

confidence: 99%

Section: Instantaneous Hand Gesture Recognition Neural Network Andmentioning

confidence: 99%

See 2 more Smart Citations

Surface EMG-Based Instantaneous Hand Gesture Recognition Using Convolutional Neural Network with the Transfer Learning Method

Zhao

Wang

et al. 2021

Sensors

View full text Add to dashboard Cite

show abstract

“…Recently, deep convolutional neural networks (CNNs) have achieved tremendous successes in a broad range of applications, such as speech recognition, gesture recognition, and natural language processing [17][18][19][20]. Their considerable feature extracting power also contributes to their success in HSI classification.…”

Section: Introductionmentioning

confidence: 99%

Triple-Attention-Based Parallel Network for Hyperspectral Image Classification

Zhu

Zheng

et al. 2021

Remote Sensing

View full text Add to dashboard Cite

Convolutional neural networks have been highly successful in hyperspectral image classification owing to their unique feature expression ability. However, the traditional data partitioning strategy in tandem with patch-wise classification may lead to information leakage and result in overoptimistic experimental insights. In this paper, we propose a novel data partitioning scheme and a triple-attention parallel network (TAP-Net) to enhance the performance of HSI classification without information leakage. The dataset partitioning strategy is simple yet effective to avoid overfitting, and allows fair comparison of various algorithms, particularly in the case of limited annotated data. In contrast to classical encoder–decoder models, the proposed TAP-Net utilizes parallel subnetworks with the same spatial resolution and repeatedly reuses high-level feature maps of preceding subnetworks to refine the segmentation map. In addition, a channel–spectral–spatial-attention module is proposed to optimize the information transmission between different subnetworks. Experiments were conducted on three benchmark hyperspectral datasets, and the results demonstrate that the proposed method outperforms state-of-the-art methods with the overall accuracy of 90.31%, 91.64%, and 81.35% and the average accuracy of 93.18%, 87.45%, and 78.85% over Salinas Valley, Pavia University and Indian Pines dataset, respectively. It illustrates that the proposed TAP-Net is able to effectively exploit the spatial–spectral information to ensure high performance.

show abstract

Intelligent Systems for Muscle Tracking: A Review on Sensor‐Algorithm Synergy

Putcha

Nguyen

Smith

et al. 2023

Advanced Intelligent Systems

View full text Add to dashboard Cite

Advanced technologies for muscle tracking provide easy access to identify and track muscle activity, often for the purposes of therapeutic interventions. The necessity of muscle trackers arises from the acute and chronic sources that disrupt neuromuscular control, resulting in an impaired ability to perform daily activities without assistance. In the context of human–machine interfaces, muscle trackers can serve as the “sensory” component, providing real‐time information to machines, such as exoskeletons and prosthetics, that can act upon such information for therapeutic and functional aid. Recently developed devices for muscle tracking rely on combinations of sensor modalities and algorithms to extract information from biosignals track muscle activity, or even extrapolate kinematic information, including gestures. However, a number of obstacles remain to be overcome to further facilitate the practical implementation of muscle‐tracking technologies, the most notable being real‐time analysis of biosignal data and extracting kinematic information from complex movements. This review attempts to cover the mechanisms behind various sensor modalities and algorithms commonly used for muscle tracking, as well as establish the current state of applications within the field. Given its multidisciplinary nature and ability to free users from rehabilitation constraints, the field of muscle tracking holds significant promise for future study.

show abstract

Hand Gesture Recognition based on Surface Electromyography using Convolutional Neural Network with Transfer Learning Method

Cited by 126 publications

References 24 publications

Surface EMG-Based Instantaneous Hand Gesture Recognition Using Convolutional Neural Network with the Transfer Learning Method

Surface EMG-Based Instantaneous Hand Gesture Recognition Using Convolutional Neural Network with the Transfer Learning Method

Triple-Attention-Based Parallel Network for Hyperspectral Image Classification

Intelligent Systems for Muscle Tracking: A Review on Sensor‐Algorithm Synergy

Contact Info

Product

Resources

About