Paper Abstract and Keywords |
Presentation |
2021-12-15 12:25
Enhancement of sign language motion classification accuracy by adding finger information using OpenPose Tsukasa Wakao, Wataru Odagiri, Sato Tatsuya, Yuusuke Kawakita, Hiromitsu Nishimura, Hiroshi Tanaka (KAIT) |
Abstract |
(in Japanese) |
(See Japanese page) |
(in English) |
As a sign language motion classification method, the direction vector from neck to shoulder, elbow, and wrist was calculated by using the nodal position information of the sign language motion using OpenPose. Each element was normalized by distance from neck to shoulder, and the data and SVM were used to create a training model for classifying each motion. In this study, we describe the results of classification accuracy enhancement by adding nodal position information of five fingers. |
Keyword |
(in Japanese) |
(See Japanese page) |
(in English) |
sign language motion classification / OpenPose / support vector machine (SVM) / finger information / / / / |
Reference Info. |
IEICE Tech. Rep. |
Paper # |
|
Date of Issue |
|
ISSN |
|
Download PDF |
|