ETRI-Knowledge Sharing Plaform

KOREAN
논문 검색
Type SCI
Year ~ Keyword

Detail

Conference Paper Elaborated real-time 3D human pose estimation on simulator
Cited - time in scopus Share share facebook twitter linkedin kakaostory
Authors
MINJONG KIM, SANGSEUNG KANG, SUYOUNG CHI
Issue Date
2018-08
Citation
International Conference on Big Data Applications and Services (BigDAS) 2018, pp.361-368
Language
English
Type
Conference Paper
Abstract
This paper presents an accurate real-time human pose estimation using a single depth image on a simulator. Our algorithm estimates seven sides of the human skeleton within an approximately ten-pixel difference compared with those of the ground truth dataset. For the proposed human pose estimation, we combine two methods: random tree walk with a depth comparison, and pixelwise classification within a local area. To evaluate our proposed algorithm, we used the ETRI database. Our evaluation on a simulator shows very sophisticated accuracy within approximately a ten-pixel difference at a speed of 40 fps on the simulator.
KSP Keywords
3D human pose estimation, Ground Truth, Human Skeleton, Random tree, Real-time 3d, Single Depth Image, local area, pixel difference