| Title |
Design of Finger Sign Language Classification using Convolutional Neural Networks |
| Authors |
김기범(Gibeom Kim) ; 이상윤(Sangyoon Lee) ; 윤창용(Changyong Yoon) ; 홍성준(Sungjun Hong) |
| DOI |
https://doi.org/10.5370/KIEE.2022.71.10.1405 |
| Keywords |
Finger sign language recognition; Gesture recognition; Convolutional neural networks; LeNet-5 |
| Abstract |
In the post-COVID-19 era, remote education and discussion using video conferencing applications are spreading, but hearing-impaired people have difficulties in interactive. In this paper, we design a finger sign language recognition system using convolutional neural networks to solve the difficulties experienced by the hearing impaired. The hand area is detected from the image acquired by a camera using the Google MediaPipe Hands solution, and the detected hand area is classified by finger sign language classification model based on LeNet-5. In order to evaluate the performance of the finger sign language classification, the American finger sign language dataset (AFSL dataset) consisting of Alphabet sign language and numeric sign language is constructed from open dataset and the feasibility of real-time finger sign language recognition system is confirmed by implementing a prototype. |