BEGIN:VCALENDAR
VERSION:2.0
PRODID:Linklings LLC
BEGIN:VTIMEZONE
TZID:Asia/Seoul
X-LIC-LOCATION:Asia/Seoul
BEGIN:STANDARD
TZOFFSETFROM:+0900
TZOFFSETTO:+0900
TZNAME:KST
DTSTART:18871231T000000
DTSTART:19881009T020000
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
DTSTAMP:20230103T035309Z
LOCATION:Room 324\, Level 3\, West Wing
DTSTART;TZID=Asia/Seoul:20221207T140000
DTEND;TZID=Asia/Seoul:20221207T153000
UID:siggraphasia_SIGGRAPH Asia 2022_sess159_papers_130@linklings.com
SUMMARY:Disentangled Image Colorization via Global Anchors
DESCRIPTION:Technical Communications, Technical Papers\n\nDisentangled Ima
 ge Colorization via Global Anchors\n\nXia, Hu, Wong, Wang\n\nColorization 
 is multimodal by nature and challenges existing frameworks to achieve colo
 rful and structurally consistent results. Even the sophisticated autoregre
 ssive model struggles to maintain long-distance color consistency due to t
 he fragility of sequential dependence.\nTo overcome this challenge, we pro
 pose a novel colorization framework that disentangles color multimodality 
 and structure consistency through global color anchors, so that both aspec
 ts could be learned effectively. Our key insight is that several carefully
  located anchors could represent the color distribution of an image approx
 imately, and conditioned on the anchor colors, we can predict the image co
 lor in a deterministic manner by leveraging the internal correlation. To t
 his end, we construct a colorization model with dual branches, where the c
 olor modeler predicts the color distribution for anchor color representati
 on, and the color generator predicts the pixel colors by referring the sam
 pled anchor colors. Importantly, the anchors are located under two princip
 les: color independence and global coverage, which is realized with cluste
 ring analysis on the deep color features. To simplify the computation, we 
 creatively adopt soft superpixel segmentation to reduce the image primitiv
 es, which still nicely reserves the reversibility to pixel-wise representa
 tion. Extensive experiments show that our method achieves notable superior
 ity over various mainstream frameworks in perceptual quality. Thanks to an
 chor-based color representation, our model has the flexibility to support 
 diverse and controllable colorization as well.\n\nRegistration Category: F
 ULL ACCESS, ON-DEMAND ACCESS\n\nLanguage: ENGLISH\n\nFormat: IN-PERSON, ON
 -DEMAND
URL:https://sa2022.siggraph.org/en/full-program/?id=papers_130&sess=sess15
 9
END:VEVENT
END:VCALENDAR