BEGIN:VCALENDAR
VERSION:2.0
PRODID:Linklings LLC
BEGIN:VTIMEZONE
TZID:Asia/Tokyo
X-LIC-LOCATION:Asia/Tokyo
BEGIN:STANDARD
TZOFFSETFROM:+0900
TZOFFSETTO:+0900
TZNAME:JST
DTSTART:18871231T000000
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
DTSTAMP:20250110T023313Z
LOCATION:Hall B7 (1)\, B Block\, Level 7
DTSTART;TZID=Asia/Tokyo:20241206T150800
DTEND;TZID=Asia/Tokyo:20241206T151900
UID:siggraphasia_SIGGRAPH Asia 2024_sess150_papers_189@linklings.com
SUMMARY:World-Grounded Human Motion Recovery via Gravity-View Coordinates
DESCRIPTION:Technical Papers\n\nZehong Shen, Huaijin Pi, Yan Xia, Zhi Cen,
  and Sida Peng (State Key Laboratory of CAD&CG, Zhejiang University); Zech
 en Hu (Deep Glint); Hujun Bao (State Key Laboratory of CAD&CG, Zhejiang Un
 iversity); Ruizhen Hu (Shenzhen University (SZU)); and Xiaowei Zhou (State
  Key Laboratory of CAD&CG, Zhejiang University)\n\nWe present a novel meth
 od for recovering world-grounded human motion from monocular video. The ma
 in challenge lies in the ambiguity of defining the world coordinate system
 , which varies between sequences. Previous approaches attempt to alleviate
  this issue by predicting relative motion in an autoregressive manner, but
  are prone to accumulating errors. Instead, we propose estimating human po
 ses in a novel Gravity-View (GV) coordinate system, which is defined by th
 e world gravity and the camera view direction. The proposed GV system is n
 aturally gravity-aligned and uniquely defined for each video frame, largel
 y reducing the ambiguity of learning image-pose mapping. The estimated pos
 es can be transformed back to the world coordinate system using camera rot
 ations, forming a global motion sequence. Additionally, the per-frame esti
 mation avoids error accumulation in the autoregressive methods. Experiment
 s on in-the-wild benchmarks demonstrate that our method recovers more real
 istic motion in both the camera space and world-grounded settings, outperf
 orming state-of-the-art methods in both accuracy and speed. The code is av
 ailable at https://zju3dv.github.io/gvhmr.\n\nRegistration Category: Full 
 Access, Full Access Supporter\n\nLanguage Format: English Language\n\nSess
 ion Chair: Li-Yi Wei (Adobe Research)
URL:https://asia.siggraph.org/2024/program/?id=papers_189&sess=sess150
END:VEVENT
END:VCALENDAR