Article
Engineering, Biomedical
Tiantong Wang, Yunbiao Zhao, Qining Wang
Summary: This paper introduces a flexible wristband and utilizes a triplet network for inter-day hand gesture recognition. The results show that the system achieves high recognition accuracy and outperforms the convolutional neural network trained with softmax-cross-entropy loss. Additionally, the size of the capacitive array has an impact on the inter-day classification result.
IEEE TRANSACTIONS ON NEURAL SYSTEMS AND REHABILITATION ENGINEERING
(2022)
Article
Engineering, Electrical & Electronic
Guan Yuan, Xiao Liu, Qiuyan Yan, Shaojie Qiao, Zhixiao Wang, Li Yuan
Summary: This study introduces a novel data glove and proposes an improved deep feature fusion network, achieving good results in gesture recognition, especially in recognizing American Sign Language and Chinese Sign Language.
IEEE SENSORS JOURNAL
(2021)
Article
Engineering, Electrical & Electronic
Zhaocheng Yang, Xinbo Zheng
Summary: The paper introduces a touchless hand gesture recognition method using radar sensor, including range-Doppler-angle trajectories extraction and a reused LSTM network. This method utilizes a 77GHz radar and a gesture desktop to improve recognition accuracy.
IEEE SENSORS JOURNAL
(2021)
Article
Computer Science, Information Systems
Hong Wang, Peiqi Kang, Qinghua Gao, Shuo Jiang, Peter B. Shull
Summary: Wrist-based hand gesture recognition has great potential for virtual and augmented reality applications. This study proposes a multi-modal sensing approach that combines photoplethysmography, force myography, and accelerometry for gesture recognition using a convolutional neural network. Experimental results show that the multi-modal fusion approach significantly improves classification performance compared to single sensing modalities.
IEEE JOURNAL OF BIOMEDICAL AND HEALTH INFORMATICS
(2022)
Article
Computer Science, Information Systems
Jianyong Li, Chengbei Li, Jihui Han, Yuefeng Shi, Guibin Bian, Shuai Zhou
Summary: The proposed method in this study achieves high accuracy in recognizing multi-scale and multi-angle hand gestures against complex backgrounds through feature extraction and SVM classification.
Article
Computer Science, Artificial Intelligence
Michalis Lazarou, Bo Li, Tania Stathaki
Summary: This work introduces a novel shape matching methodology for real-time hand gesture recognition and demonstrates its superiority in accuracy and computational efficiency through comparison with other methods.
COMPUTER VISION AND IMAGE UNDERSTANDING
(2021)
Article
Computer Science, Artificial Intelligence
Sakshi Sharma, Sukhwinder Singh
Summary: Hand gestures are crucial for communication and form the foundation of sign language, which is a visual form of communication. A deep learning CNN model designed for recognizing gesture-based sign language achieved high classification accuracy with fewer model parameters. The proposed model outperformed existing techniques in classifying gestures accurately with minimal error rates.
EXPERT SYSTEMS WITH APPLICATIONS
(2021)
Article
Chemistry, Analytical
Shenglin Zhao, Haoyuan Cai, Wenkuan Li, Yaqian Liu, Chunxiu Liu
Summary: The paper introduces a low-cost hand gesture recognition algorithm that identifies gestures through processing three-axis linear acceleration and axis-crossing codes, with high accuracy and low time costs. This algorithm is competitive in the field of consumer electronics.
Article
Computer Science, Information Systems
Abir Sen, Tapas Kumar Mishra, Ratnakar Dash
Summary: Hand gesture recognition, as an alternative for human-machine interaction, has been widely used in various fields such as 3D game technology, sign language interpreting, VR environment, and robotics. This paper presents an ensemble of CNN-based approaches to overcome the limitations of CNN architectures, including high variance, overfitting, and prediction errors. Experimental results show that the proposed ensemble model outperforms existing state-of-the-art approaches.
MULTIMEDIA TOOLS AND APPLICATIONS
(2022)
Article
Computer Science, Artificial Intelligence
ZhenZhou Wang
Summary: This paper proposes a hand gesture recognition method based on the contour features extracted by slope difference distribution (SDD), achieving 100% recognition accuracy on multiple public datasets.
APPLIED INTELLIGENCE
(2022)
Article
Multidisciplinary Sciences
Ramin Fathian, Steven Phan, Chester Ho, Hossein Rouhani
Summary: This study aimed to develop and validate a technology using a wrist-mounted inertial measurement unit (IMU) based on dynamic time warping (DTW) and k-nearest neighbours (KNN) for detecting and monitoring face touch in controlled and natural environment trials. The sensitivity, precision, and accuracy of the developed technology were evaluated, showing high performance in both controlled and natural environment trials. In conclusion, the wrist-mounted IMU can be used as an ambulatory system to detect and monitor face touching as a high-risk habit in daily life.
Article
Engineering, Electrical & Electronic
Shengcai Duan, Le Wu, Bo Xue, Aiping Liu, Ruobing Qian, Xun Chen
Summary: In this study, a novel hybrid fusion model called HyFusion is proposed for hand gesture recognition based on sEMG and ACC. It effectively combines multimodal features using multiscale attention and metric learning, leading to improved performance.
IEEE SENSORS JOURNAL
(2023)
Article
Computer Science, Information Systems
Tuan Linh Dang, Trung Hieu Pham, Quang Minh Dang, Nicolas Monet
Summary: This paper proposes a lightweight architecture for recognizing hand gestures on resource-constrained devices. The architecture consists of two main components: a segmentation algorithm for preprocessing to remove noise and irrelevant parts, and a classification algorithm for gesture recognition. Different lightweight segmentation and classification algorithms were investigated and customized. Experimental results demonstrate that the proposed architecture achieves high accuracy with various datasets, even in the presence of noise and complicated backgrounds, particularly with the combinations of DeepLabV3+ for segmentation and MobileNetV2 or EfficientNetB0 for classification. Additionally, the lightweight system achieves an inference speed of approximately 20 milliseconds with the fastest backbone, without the need for a high-end GPU.
MULTIMEDIA TOOLS AND APPLICATIONS
(2023)
Article
Computer Science, Information Systems
Jun Ma, Xunhuan Ren, Hao Li, Wenzu Li, Viktar Yurevich Tsviatkou, Anatoliy Antonovich Boriskevich
Summary: This paper proposes a skeleton extraction framework to enhance the robustness of existing skeletonization methods against both inner and border noise. By using different scales of Gaussian filters to smooth the input image and obtaining multiple representations, followed by binarization and skeletonization to produce a series of binary and skeletal images, the most suitable skeleton is selected based on a measurement that considers both the changes in skeleton and binary images. Experimental results show that the proposed framework can reduce inner noise by approximately 92% and border noise by approximately 40% based on the measure of skeleton variation rate.
Article
Automation & Control Systems
Tse-Yu Pan, Wan-Lun Tsai, Chen-Yuan Chang, Chung-Wei Yeh, Min-Chun Hu
Summary: This study proposes a training system for sports referees that utilizes deep belief networks to learn gesture features and achieve robust recognition results by combining them with selective handcrafted features. It also introduces a hierarchical recognition scheme and fusion of multimodality data to improve recognition accuracy.
IEEE TRANSACTIONS ON CYBERNETICS
(2022)