Learning to Factorize and Relight a City
Learning to Factorize and Relight a City
复制标题
学习分解和重新点亮城市
DOI：
10.1007/978-3-030-58548-8_32
复制
发表时间：
2020
期刊：
影响因子：
--
通讯作者：
Snavely, Noah
中科院分区：
文献类型：
--
作者：
Liu, Andrew;Ginosar, Shiry;Zhou, Tinghui;Efros, Alexei;Snavely, Noah
我们提出了一种基于学习的框架，用于将室外场景分解为随时间变化的照明和永久场景因素。受经典的内在图像分解的启发，我们的学习信号建立在两个见解的基础上：1）结合解开的因素应该重建原始图像，2）永久因素应该在同一场景的多个时间样本中保持不变。为了便于训练，我们收集了来自谷歌街景的城市规模的室外延时图像数据集，其中随着时间的推移重复捕获相同的位置。这些数据代表了前所未有的时空户外图像规模。我们证明，我们学到的解开因素可以用来以现实的方式操纵新颖的图像，例如改变灯光效果和场景几何形状。请访问 http://factorize-a-city.github.io/ 查看动画结果。
We propose a learning-based framework for disentangling outdoor scenes into temporally-varying illumination and permanent scene factors. Inspired by the classic intrinsic image decomposition, our learning signal builds upon two insights: 1) combining the disentangled factors should reconstruct the original image, and 2) the permanent factors should stay constant across multiple temporal samples of the same scene. To facilitate training, we assemble a city-scale dataset of outdoor timelapse imagery from Google Street View, where the same locations are captured repeatedly through time. This data represents an unprecedented scale of spatio-temporal outdoor imagery. We show that our learned disentangled factors can be used to manipulate novel images in realistic ways, such as changing lighting effects and scene geometry. Please visit http://factorize-a-city.github.io/ for animated results.
DOI：
--
发表时间：
2017-11
期刊：
ArXiv
影响因子：
--
作者：
Michael Janner;Jiajun Wu;Tejas D. Kulkarni;Ilker Yildirim;J. Tenenbaum
通讯作者：
Michael Janner;Jiajun Wu;Tejas D. Kulkarni;Ilker Yildirim;J. Tenenbaum
DOI：
--
发表时间：
2011
期刊：
Computer Vision and Pattern Recognition
影响因子：
--
作者：
Michael Rubinstein;Ce Liu;Peter Sand;F. Durand;W. Freeman
通讯作者：
W. Freeman