BEGIN:VCALENDAR
VERSION:2.0
PRODID:Linklings LLC
BEGIN:VTIMEZONE
TZID:Asia/Hong_Kong
X-LIC-LOCATION:Asia/Hong_Kong
BEGIN:STANDARD
TZOFFSETFROM:+0800
TZOFFSETTO:+0800
TZNAME:HKT
DTSTART:19911015T033000
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
DTSTAMP:20251218T030656Z
LOCATION:Meeting Room S421\, Level 4
DTSTART;TZID=Asia/Hong_Kong:20251215T133100
DTEND;TZID=Asia/Hong_Kong:20251215T134200
UID:siggraphasia_SIGGRAPH Asia 2025_sess104_papers_2078@linklings.com
SUMMARY:Surface-Aware Distilled 3D Semantic Features
DESCRIPTION:Lukas Uzolas, Elmar Eisemann, and Petr Kellnhofer (Delft Unive
 rsity of Technology)\n\nMany 3D tasks such as pose alignment, animation, m
 otion transfer, and 3D reconstruction rely on establishing correspondences
  between 3D shapes. This challenge has recently been approached by pairwis
 e matching of semantic features from pre-trained vision models. However, d
 espite their power, these features struggle to differentiate instances of 
 the same semantic class such as ``left hand'' versus ``right hand'' which 
 leads to substantial mapping errors. To solve this, we learn a surface-awa
 re embedding space that is robust to these ambiguities while facilitating 
 shared mapping for an entire family of 3D shapes. Importantly, our approac
 h is self-supervised and requires only a small number of unpaired training
  meshes to infer features for new possibly imperfect 3D shapes at test tim
 e. We achieve this by introducing a contrastive loss that preserves the se
 mantic content of the features distilled from foundational models while di
 sambiguating features located far apart on the shape's surface. We observe
  superior performance in correspondence matching benchmarks and enable dow
 nstream applications including 2D-to-3D and 3D-to-3D texture transfer, in-
 part segmentation, pose alignment, and motion transfer in low-data regimes
 . Unlike previous pairwise approaches, our solution constructs a joint emb
 edding space, where both seen and unseen 3D shapes are implicitly aligned 
 without further optimization.\n\nRegistration Category: Full Access, Full 
 Access Supporter\n\nSession Chair: Xuejin Chen (University of Science and 
 Technology of China)\n\n
END:VEVENT
END:VCALENDAR