SfM with MRFs: Discrete-Continuous Optimization for Large-Scale Structure from Motion

By David J. Crandall, Andrew Owens, Noah Snavely, Dan Huttenlocher. 2013

390 citations recorded by the source corpus.

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

Citations in the corpus

DUSt3R: Geometric 3D Vision Made Easy (2023, background): “Structure-from-Motion (SfM) [20, 21, 45, 48, 106] aims at reconstructing sparse 3D maps while jointly determining camera parameters from a set of images.”
Comparative Evaluation of Hand-Crafted and Learned Local Features (2017, unknown): “[8].”
From single image query to detailed 3D reconstruction (2015, methodology): “[5] proposed a global method that performs SfM based on a MRF optimization.”
First-person hyper-lapse videos (2014, methodology): “In order to compute our 3D camera motions and scene proxies, we build upon techniques from structure-from-motion, which has seen dramatic progress in recent years [Snavely et al. 2006; Crandall et al. 2013; Wu 2013].”
Reconstructing the World* in Six Days *(As Captured by the Yahoo 100 Million Image Dataset) (2015, methodology): “[8] propose a novel LS-SfM technique for geo-localized images by leveraging a Markov Random Field.”
Fine-grained semi-supervised labeling of large shape collections (2013, methodology): “Despite some recent advances [Huber 2002; Crandall et al. 2011; Kim et al. 2012; Huang et al. 2012; Kim et al. 2013] on this topic, we found that it is extremely difficult to obtain high-quality correspondences among a shape collection of thousands of shapes.”
Predicting Geo-informative Attributes in Large-Scale Image Collections Using Convolutional Neural Networks (2015, methodology): “These techniques work well for popular landmarks, but cannot infer geo-tags for photos taken outside these places.”
Robust reconstruction of indoor scenes (2015, background): “In structure from motion estimation, robustness can be increased using appropriate penalty functions [12, 27, 8] or by identifying inconsistent substructures among pairwise relations between camera poses [67, 68, 50, 63].”
Where am I in the dark: Exploring active transfer learning on the use of indoor localization based on thermal imaging (2016, methodology): “3D SfM models [6,49] are used for image-based localization problems in enhancing the accuracy.”
Learned Contextual Feature Reweighting for Image Geo-Localization (2017, background): “Visual image geo-localization has been an active research area for the past decade [2, 28, 48, 53], owing to its wide range of applications including augmented reality [36], autonomous driving [33], adding and refining geo-tags in image collections [19, 65], large-scale 3D reconstruction [11], and p”
Robust Global Translations with 1DSfM (2014, methodology): “Following a now-common approach [11,16,17,6,3,14], we separate the initialization into two stages: a rotations problem and a translations problem.”