TCCNN: Transformer ConCated Convolutional Neural Networks for Hand Gesture Recognition
Konferenz: CAIBDA 2022 - 2nd International Conference on Artificial Intelligence, Big Data and Algorithms
17.06.2022 - 19.06.2022 in Nanjing, China
Tagungsband: CAIBDA 2022
Seiten: 3Sprache: EnglischTyp: PDF
Autoren:
Geng, Jianchun (Department of Information Engineering, Wei Fang Engineering Vocational College, Weifang, China)
Wen, Lili (College of Teacher Education, Wei Fang Engineering Vocational College, Weifang, China)
Geng, Jiannuan (Library, Hebei University of Science and Technology, Shijiazhuang, China)
Inhalt:
Gesture recognition has important significance to human-computer interaction. The machine vision based gesture recognition has a wide range of application scenarios since its convenience and low cost. A deep learning model which combined convolutional neural networks (CNN) and Transformer is proposed in this paper. Local features extracted by CNN and long-range dependencies between patches captured by Transformer are both taken into account for gesture recognition. The proposed model was trained and tested on OUTHANDS and HGR1 datasets. The recognition accuracy of this model was proved to be much higher compared with other existing methods.