Modeling the World from Internet Photo Collections

By Noah Snavely, S. M. Seitz, Richard Szeliski. International Journal of Computer Vision, 2008

2,410 citations recorded by the source corpus.

Subfields: Computer Vision and Pattern Recognition

View Paper: https://www.semanticscholar.org/paper/5e70bb805cb99904504f12b3fa32ddaae066323c

How researchers see this paper

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

In Computer Vision and Pattern Recognition

Citations in the corpus

Building Rome on a Cloudless Day (2010, background): “Sparse 3D reconstruction of landmarks from Internet photo collections was first addressed by the Photo Tourism system [20], which achieves high-quality results through exhaustive pairwise image matching and frequent global bundle adjustment.”
Parallel Poisson Surface Reconstruction (2009, background): “New computer vision techniques [2] allow three dimensional point clouds to be extracted from photo collections; with an abundance of photographs of the same scene available through online photo sites, the potential for truly massive datasets is within reach.”
Assessing the Accuracy of Georeferenced Point Clouds Produced via Multi-View Stereopsis from Unmanned Aerial Vehicle (UAV) Imagery (2012, methodology): “[22] whereby a least squares bundle adjustment is performed based on the matching of SIFT features from down-sampled versions of the images.”
Pegasus: Coordinated Scheduling for Virtualized Accelerator-based Systems (2011, background): “An example is the execution of sequences of operations like those that first identify spatial correlation or correspondence [33] between images prior to synthesizing them [25].”
A Simple Prior-Free Method for Non-rigid Structure-from-Motion Factorization (2012, methodology): “Most structure-from-motion techniques used for rigid scene reconstruction (e.g., Pollefeys et al. 2004; Snavely et al. 2008; Dai et al. 2010; Li 2010) rely on feature point correspondences.”
Evaluation of image-based modeling and laser scanning accuracy for emerging automated performance monitoring techniques (2011, methodology): “, [15], [36]) to start with a good initial image pair and good estimates for camera parameters in the automatically chosen pair.”
Forest Structure Estimation from a UAV-Based Photogrammetric Point Cloud in Managed Temperate Coniferous Forests (2017, methodology): “Recent advances in computer science make it possible to generate a point cloud semi-automatically using the Structure from Motion (SfM) approach [7,8].”
Unmanned aerial systems for photogrammetry and remote sensing: A review (2014, background): “On the contrary, the irregularity of UAS blocks was no obstacle to automatic image matching and bundle adjustment software originated in the computer vision community for more general purposes—the so-called Structure from Motion (SfM) approach (Snavely et al., 2008; Agarwal et al., 2009),—or specifi”
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.”
Reconstructing the World* in Six Days *(As Captured by the Yahoo 100 Million Image Dataset) (2015, background): “[25, 26] demonstrated the feasibility of SfM for several thousand Internet images.”
Remote sensing of forest pest damage: a review and lessons learned from a Canadian perspective* (2016, methodology): “…are a sort of three-dimensional model of the canopy; can be derived from overlapping photographs taken from consumer-grade cameras using the “structure from motion” computer vision approach (Snavely et al. 2008); and have a much lower cost in data acquisition and processing than LiDAR point clouds.”
Continuous 3D Perception Model with Persistent State (2025, neutral): “Notable examples include traditional methods such as Structure from Motion (SfM) [1, 2, 28, 47, 65, 66, 70, 71, 80] and Simultaneous Localization and Mapping (SLAM) [11, 16, 19, 22, 37, 51, 53], as well as more recent approaches such as Neural Radiance Fields (NeRF) [12, 24, 49, 50, 84] and 3D…”
Unsupervised 3D object recognition and reconstruction in unordered datasets (2005, methodology): “Sample data [(2,3), (5,4), (9,6), (4,7), (8,1), (7,2)]”