If you missed SIGGRAPH 2017 watch a selection of recorded Live Streaming Sessions.
If you missed SIGGRAPH 2017 watch a selection of recorded Live Streaming Sessions.
A new CNN-based method for extracting clear and smooth structural lines from screen-rich manga.
Chengze Li
The Chinese University of Hong Kong
Xueting Liu
The Chinese University of Hong Kong
Tien-Tsin Wong
The Chinese University of Hong Kong
This study replaces a large part of image processing pipelines with a data-driven model that learns local affine models in bilateral space from image pairs. The algorithm provides a real-time viewfinder on smartphones, with state-of-the-art approximation quality, and it can even learn the edits of human retouchers.
Michael Gharbi
Massachusetts Institute of Technology
Jiawen Chen
Google Research
Jonathan Barron
Google Research
Samuel Hasinoff
Google Research
Frédo Durand
Massachusetts Institute of Technology
This paper proposes a deep learning approach for user-guided image colorization. The system directly maps a grayscale image, along with sparse, local user "hints" to an output colorization. The CNN propagates user edits by fusing low-level cues with high-level semantic information learned from large-scale data.
Richard Zhang
University of California, Berkeley
Jun-Yan Zhu
University of California, Berkeley
Phillip Isola
University of California, Berkeley
Xinyang Geng
University of California, Berkeley
Angela S. Lin
University of California, Berkeley
Yu Tianhe
University of California, Berkeley
Alexei A. Efros
University of California, Berkeley
This new technique for visual-attribute transfer across images that may have very different appearance but have perceptually similar semantic structure adapts the notion of "image analogy" with features extracted from a deep convolutional neural network to find semantically meaningful dense correspondences between two input images.
Jing Liao
Microsoft Research Asia
Yuan Yao
Shanghai Jiao Tong University
Lu Yuan
Microsoft Research Asia
Gang Hua
Microsoft Research Asia
Sing Bing Kang
Microsoft Corporation, Microsoft Research
This work applies the theory of Lie Groups, using a group whose operator is Porter & Duff's Over, to derive many methods, both standard and new, for combining, compressing, and computing with deep images.
Tom Duff
Pixar Animation Studios