TY  - CONF
AU  - Boomers, Ann Katrin
AU  - Boltes, Maik
TI  - Shoulder Rotation Measurement in Camera and 3D Motion Capturing Data
VL  - 443
CY  - Singapore
PB  - Springer Nature Singapore
M1  - FZJ-2024-03490
SN  - 978-981-99-7975-2 (print)
T2  - Lecture Notes in Civil Engineering
SP  - 35 - 42
PY  - 2024
N1  - Missing Journal: = 2366-2565 (import from CrossRef Book Series, Journals: juser.fz-juelich.de)
AB  - The individual movement of pedestrians and their body parts, as for example shoulders, is of great interest to understand body movement and interactions and thus to improve pedestrian models. Nearly all laboratory experiments in pedestrian dynamics use camera data to obtain trajectories. A perpendicular top view of the camera does not only allow to extract the head position but also data of upper body segments. The detection is more reliable if shoulders are tagged with markers and for low densities of people. In this study a head-shoulder model is used to assign coloured shoulder markers to a person. The location of a marker is predicted by taking head position, basic body dimensions, movement direction and camera angle into account. It is implemented as a new feature in the software PeTrack. This paper shows a comparison of shoulder rotation measurements obtained from 3D motion capturing systems (Xsens) with those from camera data using the newly introduced model and detection technique. Detection rates and limits of the camera-based rotation measurement are shown and implications are given for the future application at high densities in crowds.
T2  - Traffic and Granular Flow '22
CY  - 15 Oct 2022 - 17 Oct 2022, New Delhi (India)
Y2  - 15 Oct 2022 - 17 Oct 2022
M2  - New Delhi, India
LB  - PUB:(DE-HGF)8 ; PUB:(DE-HGF)7
UR  - <Go to ISI:>//WOS:001283974700005
DO  - DOI:10.1007/978-981-99-7976-9_5
UR  - https://juser.fz-juelich.de/record/1026672
ER  -