Single Image 3D Without a Single 3D Image - Robotics Institute Carnegie Mellon University

Single Image 3D Without a Single 3D Image

David F. Fouhey, Abhinav Gupta, and Martial Hebert
Conference Paper, Proceedings of (ICCV) International Conference on Computer Vision, pp. 1053 - 1061, December, 2015

Abstract

Do we really need 3D labels in order to learn how to predict 3D? In this paper, we show that one can learn a mapping from appearance to 3D properties without ever seeing a single explicit 3D label. Rather than use explicit supervision, we use the regularity of indoor scenes to learn the mapping in a completely unsupervised manner. We demonstrate this on both a standard 3D scene understanding dataset as well as Internet images for which 3D is unavailable, precluding supervised learning. Despite never seeing a 3D label, our method produces competitive results.

BibTeX

@conference{-2015-113346,
author = {David F. Fouhey and Abhinav Gupta and Martial Hebert},
title = {Single Image 3D Without a Single 3D Image},
booktitle = {Proceedings of (ICCV) International Conference on Computer Vision},
year = {2015},
month = {December},
pages = {1053 - 1061},
}