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:20260114T163656Z
LOCATION:Meeting Room C4.11\, Level 4 (Convention Centre)
DTSTART;TZID=Australia/Melbourne:20231215T145500
DTEND;TZID=Australia/Melbourne:20231215T150500
UID:siggraphasia_SIGGRAPH Asia 2023_sess138_papers_193@linklings.com
SUMMARY:Self-Calibrating, Fully Differentiable NLOS Inverse Rendering
DESCRIPTION:Kiseok Choi, Inchul Kim, and Dongyoung Choi (Korea Advanced In
 stitute of Science and Technology (KAIST)); Julio Marco and Diego Gutierre
 z (Universidad de Zaragoza - I3A); and Min H. Kim (Korea Advanced Institut
 e of Science and Technology (KAIST))\n\nExisting time-resolved non-line-of
 -sight (NLOS) imaging methods reconstruct hidden scenes by inverting the o
 ptical paths of indirect illumination measured at visible relay surfaces. 
 These methods are prone to reconstruction artifacts due to inversion ambig
 uities and capture noise, which are typically mitigated through the manual
  selection of filtering functions and parameters. We introduce a fully-dif
 ferentiable end-to-end NLOS inverse rendering pipeline that self-calibrate
 s the imaging parameters during the reconstruction of hidden scenes, using
  as input only the measured illumination while working both in the time an
 d frequency domains. Our pipeline extracts a geometric representation of t
 he hidden scene from NLOS volumetric intensities and estimates the time-re
 solved illumination at the relay wall produced by such geometric informati
 on using differentiable transient rendering. We then use gradient descent 
 to optimize imaging parameters by minimizing the error between our simulat
 ed time-resolved illumination and the measured illumination. To make our p
 ipeline efficient and differentiable, we combine diffraction-based imaging
  with path-space light transport and a simple ray marching technique for s
 urface extraction. Unlike the majority of previous works, our method extra
 cts detailed, dense sets of surface points and normals of the hidden scene
 . Our results demonstrate the robustness of our method to consistently rec
 onstruct geometry and albedo, even with significant noise interference.\n\
 nRegistration Category: Full Access\n\nSession Chair: Seung-Hwan Baek (POS
 TECH)\n\n
URL:https://asia.siggraph.org/2023/full-program?id=papers_193&sess=sess138
END:VEVENT
END:VCALENDAR