BROWSE
Detail
Conference Paper
OnTV: Interactive AR Fairy Tale Application using Human Body Pose Estimation through Mobile Device
Cited - time in 
Share 
Share
- Authors
- Byunggyu Lee, Juyoung Ki, Sung-Uk Jung
- Issue Date
- 2022-02
- Citation
- International Workshop on Frontiers of Computer Vision (IW-FCV) 2022, pp.1-5
- Language
- English
- Type
- Conference Paper
- Abstract
- In order for users and virtual objects to interact in existing AR contents, special sensor such as a depth camera are required for user’s pose estimation. To solve this problem, it is necessary to estimate the user’s pose with general-purpose sensor such as a RGB camera on mobile device. In this paper, we propose an interactive AR application using a deep learning-based 2D human pose estimation network. First, we designed a deep learning network based on a heatmap that estimates the location of joint coordinates and a part affinity field that shows the relationship between joints. Second, the lightweight deep learning algorithm was used to estimate the pose of human in real time in a mobile environment. Finally, we developed an interactive educational AR application that enables 6 types of interactions using this network.
- KSP Keywords
- Deep learning network, Depth camera, Human body, Human pose estimation, Interactive AR, Joint Coordinates, Learning-based, Mobile devices, RGB camera, Real-Time, deep learning(DL)