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:20260817T171531Z
LOCATION:Hall B5 (1)\, B Block\, Level 5
DTSTART;TZID=Asia/Tokyo:20241206T154100
DTEND;TZID=Asia/Tokyo:20241206T155500
UID:siggraphasia_SIGGRAPH Asia 2024_sess148_papers_437@linklings.com
SUMMARY:iSeg: Interactive 3D Segmentation via Interactive Attention
DESCRIPTION:Itai Lang, Fei Xu, and Dale Decatur (University of Chicago); S
 udarshan Babu (Toyota Technological Institute at Chicago (TTIC)); and Rana
  Hanocka (University of Chicago)\n\nWe present iSeg, a new interactive tec
 hnique for segmenting 3D shapes. Previous works have focused mainly on lev
 eraging pre-trained 2D foundation models for 3D segmentation based on text
 . However, text may be insufficient for accurately describing fine-grained
  spatial segmentations. Moreover, achieving a consistent 3D segmentation u
 sing a 2D model is challenging since occluded areas of the same semantic r
 egion may not be visible together from any 2D view. Thus, we design a segm
 entation method conditioned on fine user clicks, which operates entirely i
 n 3D. Our system accepts user clicks directly on the shape's surface, indi
 cating the inclusion or exclusion of regions from the desired shape partit
 ion. To accommodate various click settings, we propose a novel interactive
  attention module capable of processing different numbers and types of cli
 cks, enabling the training of a single unified interactive segmentation mo
 del. We apply iSeg to a myriad of shapes from different domains, demonstra
 ting its versatility and faithfulness to the user's specifications.\n\nReg
 istration Category: Full Access, Full Access Supporter\n\nLanguage Format:
  English Language\n\nSession Chair: Peng Song (Singapore University of Tec
 hnology and Design (SUTD))\n\n
URL:https://asia.siggraph.org/2024/program/?id=papers_437&sess=sess148
END:VEVENT
END:VCALENDAR
