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:20260114T163650Z
LOCATION:Meeting Room C4.8\, Level 4 (Convention Centre)
DTSTART;TZID=Australia/Melbourne:20231215T162500
DTEND;TZID=Australia/Melbourne:20231215T163500
UID:siggraphasia_SIGGRAPH Asia 2023_sess159_papers_1035@linklings.com
SUMMARY:Neural Gradient Learning and Optimization for Oriented Point Norma
 l Estimation
DESCRIPTION:Qing Li (Tsinghua University), Huifang Feng (Xiamen University
 ), Kanle Shi (Kuaishou Technology), Yi Fang (New York University), Yu-Shen
  Liu (Tsinghua University), and Zhizhong Han (Wayne State University)\n\nW
 e propose Neural Gradient Learning (NGL), a deep learning approach to lear
 n gradient vectors with consistent orientation from 3D point clouds for no
 rmal estimation. It has excellent gradient approximation properties for th
 e underlying geometry of the data. We utilize a simple neural network to p
 arameterize the objective function to produce gradients at points using a 
 global implicit representation. However, the derived gradients usually dri
 ft away from the ground-truth oriented normals due to the lack of local de
 tail descriptions. Therefore, we introduce Gradient Vector Optimization (G
 VO) to learn an angular distance field based on local plane geometry to re
 fine the coarse gradient vectors. Finally, we formulate our method with a 
 two-phase pipeline of coarse estimation followed by refinement. Moreover, 
 we integrate two weighting functions, i.e., anisotropic kernel and inlier 
 score, into the optimization to improve the robust and detail-preserving p
 erformance. Our method efficiently conducts global gradient approximation 
 while achieving better accuracy and generalization ability of local featur
 e description. This leads to a state-of-the-art normal estimator that is r
 obust to noise, outliers and point density variations. Extensive evaluatio
 ns show that our method outperforms previous works in both unoriented and 
 oriented normal estimation on widely used benchmarks. The source code and 
 pre-trained models are available at https://github.com/LeoQLi/NGLO.\n\nReg
 istration Category: Full Access\n\nSession Chair: Fei Hou (Institute of So
 ftware, Chinese Academy of Sciences; University of Chinese Academy of Scie
 nces)\n\n
URL:https://asia.siggraph.org/2023/full-program?id=papers_1035&sess=sess15
 9
END:VEVENT
END:VCALENDAR