Fast image-based localization using direct 2D-to-3D matching

By Torsten Sattler, Bastian Leibe, Leif Kobbelt. 2011

551 citations recorded by the source corpus.

Subfields: Aerospace Engineering, Computer Vision and Pattern Recognition

How researchers see this paper

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

In Aerospace Engineering

In Computer Vision and Pattern Recognition

Citations in the corpus

CPlaNet: Enhancing Image Geolocalization by Combinatorial Partitioning of Maps (2018, methodology): “Pose estimation approaches [9,18,19,20,21,22,23] match query images against 3D models of an area, and employ 2D-3D feature correspondences to identify 6-DOF query poses.”
PlaNet - Photo Geolocation with Convolutional Neural Networks (2016, methodology): “A query image is localized by establishing correspondences between its interest points and the points in the 3D model and solving the resulting perspective-n-point (PnP) problem to obtain the camera parameters [30, 31, 40].”
NetVLAD: CNN Architecture for Weakly Supervised Place Recognition (2015, background): “shelf CNN descriptors on two challenging place recognition benchmarks. 1. Introduction Visual place recognition has received a signiﬁcant amount of attention in the past years both in computer vision [4,9,10,23,33,58,59,60,61,75,76] and robotics communities [15,16,40,42,70] motivated by, e.g., appli”
VocMatch: Efficient Multiview Correspondence for Structure from Motion (2014, background): “Our way of generating tracks can be seen as a limiting case of [19], where conventional matching is performed, but only between those features that map to the same visual word of a smaller vocabulary.”
Geometry-Aware Learning of Maps for Camera Localization (2017, methodology): “Examples include 3D landmarks for general visual SLAM methods [34, 35, 42], 3D lines and patches in semi-dense SLAM methods and indoor scenes [20, 43, 56], object-level context in semantic SLAM methods [10, 45], bag of visual word features on key frames for camera relocalization [12, 46, 47].”
Seeing 3D Chairs: Exemplar Part-Based 2D-3D Alignment Using a Large Dataset of CAD Models (2014, background): “Images are typically represented using local invariant features such as SIFT [27], which work best for textured objects such as building facades [24, 31, 35].”
Image Retrieval for Image-Based Localization Revisited (2012, methodology): “Still, the memory requirements and additional computations hinder the scalability of [28] compared to [13].”
Variational End-to-End Navigation and Localization (2018, methodology): “This work relates to a vast literature in localization and mapping [7], [8], such as visual SLAM [9], [23] and place recognition [24], [25].”
Efficient Global 2D-3D Matching for Camera Localization in a Large-Scale 3D Map (2017, methodology): “In implementing the visual-vocabulary-tree search, we use a pre-trained vocabulary of 100K visual words [42], generated with approximate k-means clustering [39] of SIFT descriptors, and choose a tree branching factor at 10.”
Geometric Loss Functions for Camera Pose Regression with Deep Learning (2017, methodology): “Although our method improves significantly over the original PoseNet model, it is still yet to reach the fine grained accuracy of these methods [50, 57, 41, 29].”
Image Geo-Localization Based on MultipleNearest Neighbor Feature Matching UsingGeneralized Graphs (2014, methodology): “We evaluated the proposed framework on a new data set of 102k street view images; the experiments show it outperforms the state of the art by 10 percent.”
Image-Based Localization Using LSTMs for Structured Feature Correlation (2016, methodology): “Given a SfM model of a scene, where each 3D point is associated with the image features from which it was triangulated, one proceeds in two stages [31, 42]: (i) establishing 2D-3D matches between features extracted from the query image and 3D points in the SfM model via descriptor matching; (ii) usi”
Get Out of My Lab: Large-scale, Real-Time Visual-Inertial Localization (2015, methodology): “Yet, most large-scale localization approaches rely on directly matching the feature descriptors against the landmark descriptors without any intermediate retrieval step [25, 26, 41, 42, 49].”
Scene Coordinate Regression Forests for Camera Relocalization in RGB-D Images (2013, methodology): “Given sparse keypoint matches, pose estimation is achieved either via intermediate image retrieval [2, 8, 14] or by directly establishing 2D-3D correspondences for a robust pose optimization [19, 28].”
Where am I in the dark: Exploring active transfer learning on the use of indoor localization based on thermal imaging (2016, methodology): “[1] directly matched descriptors of 2D images to descriptors of 3D model.”
Cross-View Image Geolocalization (2013, background): “[10, 15, 21, 14] leverage such 3D information for more efficient and accurate localization.”
Privacy Preserving Image-Based Localization (2019, methodology): “…image-based localization has led to methods that are now quite robust to changes in scene appearance and illumination [4,61], scale to large scenes [43,56,58,83], and are suitable for real-time computation and mobile devices [8, 33, 35, 43–45, 57, 76] with compressed map representations [15, 21].”
Comparative Evaluation of Hand-Crafted and Learned Local Features (2017, background): “, in Structure-fromMotion (SFM) and Multi-View Stereo (MVS) [1,17,33,37, 39,40], image retrieval [31,34,47,48], and image-based localization [35, 36, 56].”