BEGIN:VCALENDAR
VERSION:2.0
PRODID:Linklings LLC
BEGIN:VTIMEZONE
TZID:Australia/Melbourne
X-LIC-LOCATION:Australia/Melbourne
BEGIN:DAYLIGHT
TZOFFSETFROM:+1000
TZOFFSETTO:+1100
TZNAME:AEDT
DTSTART:19721003T020000
RRULE:FREQ=YEARLY;BYMONTH=4;BYDAY=1SU
END:DAYLIGHT
BEGIN:STANDARD
DTSTART:19721003T020000
TZOFFSETFROM:+1100
TZOFFSETTO:+1000
TZNAME:AEST
RRULE:FREQ=YEARLY;BYMONTH=10;BYDAY=1SU
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
DTSTAMP:20240214T070248Z
LOCATION:Meeting Room C4.11\, Level 4 (Convention Centre)
DTSTART;TZID=Australia/Melbourne:20231214T144500
DTEND;TZID=Australia/Melbourne:20231214T150000
UID:siggraphasia_SIGGRAPH Asia 2023_sess130_papers_448@linklings.com
SUMMARY:MIPS-Fusion: Multi-Implicit-Submaps for Scalable and Robust Online
  Neural RGB-D Reconstruction
DESCRIPTION:Technical Papers\n\nYijie Tang (National University of Defense
  Technology (NUDT)), Jiazhao Zhang (Peking University), Zhinan Yu (Nationa
 l University of Defense Technology (NUDT)), He Wang (Peking University), a
 nd Kai Xu (National University of Defense Technology (NUDT))\n\nWe introdu
 ce MIPS-Fusion, a robust and scalable online RGB-D reconstruction method b
 ased on a novel neural implicit representation -- multi-implicit-submap. D
 ifferent from existing neural RGB-D reconstruction methods lacking either 
 flexibility with a single neural map or scalability due to extra storage o
 f feature grids, we propose a pure neural representation tackling both dif
 ficulties with a divide-and-conquer design. In our method, neural submaps 
 are incrementally allocated alongside the scanning trajectory and efficien
 tly learned with local neural bundle adjustments. The submaps can be refin
 ed individually in a back-end optimization and optimized jointly to realiz
 e submap-level loop closure. Meanwhile, we propose a hybrid tracking appro
 ach combining randomized and gradient-based pose optimizations. For the fi
 rst time, randomized optimization is made possible in neural tracking with
  several key designs to the learning process, enabling efficient and robus
 t tracking even under fast camera motions. The extensive evaluation demons
 trates that our method attains higher reconstruction quality than the stat
 e of the arts for large-scale scenes and under fast camera motions.\n\nReg
 istration Category: Full Access\n\nSession Chair: Baoquan Chen (Peking Uni
 versity)
URL:https://asia.siggraph.org/2023/full-program?id=papers_448&sess=sess130
END:VEVENT
END:VCALENDAR