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:20260817T171534Z
LOCATION:Hall B5 (2)\, B Block\, Level 5
DTSTART;TZID=Asia/Tokyo:20241205T091400
DTEND;TZID=Asia/Tokyo:20241205T092800
UID:siggraphasia_SIGGRAPH Asia 2024_sess125_papers_563@linklings.com
SUMMARY:Online Neural Denoising with Cross-Regression for Interactive Rend
 ering
DESCRIPTION:Hajin Choi (Gwangju Institute of Science and Technology); Seok
 pyo Hong (Samsung Advanced Institute of Technology); Inwoo Ha (Samsung Adv
 anced Institute of Technology, KAIST); Nahyup Kang (Samsung Advanced Insti
 tute of Technology); and Bochang Moon (Gwangju Institute of Science and Te
 chnology)\n\nGenerating a rendered image sequence through Monte Carlo ray 
 tracing is an appealing option when one aims to accurately simulate variou
 s lighting effects. Unfortunately, interactive rendering scenarios limit t
 he allowable sample size for such sampling-based light transport algorithm
 s, resulting in an unbiased but noisy image sequence. Image denoising has 
 been widely adopted as a post-sampling process to convert such noisy image
  sequences into biased but temporally stable ones. The state-of-the-art st
 rategy for interactive image denoising involves devising a deep neural net
 work and training this network via supervised learning, i.e., optimizing t
 he network parameters using training datasets that include an extensive se
 t of image pairs (noisy and ground truth images). This paper adopts the pr
 evalent approach for interactive image denoising, which relies on a neural
  network. However, instead of supervised learning, we propose a different 
 learning strategy that trains our network parameters on the fly, i.e., upd
 ating them online using runtime image sequences. To achieve our denoising 
 objective with online\nlearning, we tailor local regression to a cross-reg
 ression form that can guide robust training of our denoising neural networ
 k. We demonstrate that our denoising framework effectively reduces noise i
 n input image sequences while robustly preserving both geometric and non-g
 eometric edges, without requiring the manual effort involved in preparing 
 an external dataset.\n\nRegistration Category: Full Access, Full Access Su
 pporter\n\nLanguage Format: English Language\n\nSession Chair: Wenzel Jako
 b (École Polytechnique Féderale de Lausanne (EPFL), NVIDIA)\n\n
URL:https://asia.siggraph.org/2024/program/?id=papers_563&sess=sess125
END:VEVENT
END:VCALENDAR