Sparse-to-Dense Hypercolumn Matching for Long-Term Visual Localization

Hugo Germain, Guillaume Bourmaud, Vincent Lepetit

Research output: Chapter in Book/Report/Conference proceedingConference contributionResearchpeer-review

Abstract

We propose a novel approach to feature point matching, suitable for robust and accurate outdoor visual localization in long-term scenarios. Given a query image, we first match it against a database of registered reference images, using recent retrieval techniques. This gives us a first estimate of the camera pose. To refine this estimate, like previous approaches, we match 2D points across the query image and the retrieved reference image. This step, however, is prone to fail as it is still very difficult to detect and match sparse feature points across images captured in potentially very different conditions. Our key contribution is to show that we need to extract sparse feature points only in the retrieved reference image: We then search for the corresponding 2D locations in the query image exhaustively. This search can be performed efficiently using convolutional operations, and robustly by using hypercolumn descriptors …
Original languageEnglish
Title of host publication2019 International Conference on 3D Vision (3DV)
Pages513-523
Publication statusPublished - 2019

Fingerprint

Cameras

Cite this

Germain, H., Bourmaud, G., & Lepetit, V. (2019). Sparse-to-Dense Hypercolumn Matching for Long-Term Visual Localization. In 2019 International Conference on 3D Vision (3DV) (pp. 513-523)

Sparse-to-Dense Hypercolumn Matching for Long-Term Visual Localization. / Germain, Hugo; Bourmaud, Guillaume; Lepetit, Vincent.

2019 International Conference on 3D Vision (3DV). 2019. p. 513-523.

Research output: Chapter in Book/Report/Conference proceedingConference contributionResearchpeer-review

Germain, H, Bourmaud, G & Lepetit, V 2019, Sparse-to-Dense Hypercolumn Matching for Long-Term Visual Localization. in 2019 International Conference on 3D Vision (3DV). pp. 513-523.
Germain H, Bourmaud G, Lepetit V. Sparse-to-Dense Hypercolumn Matching for Long-Term Visual Localization. In 2019 International Conference on 3D Vision (3DV). 2019. p. 513-523
Germain, Hugo ; Bourmaud, Guillaume ; Lepetit, Vincent. / Sparse-to-Dense Hypercolumn Matching for Long-Term Visual Localization. 2019 International Conference on 3D Vision (3DV). 2019. pp. 513-523
@inproceedings{9d4938d08dbd47178fec1b60fec6cf5e,
title = "Sparse-to-Dense Hypercolumn Matching for Long-Term Visual Localization",
abstract = "We propose a novel approach to feature point matching, suitable for robust and accurate outdoor visual localization in long-term scenarios. Given a query image, we first match it against a database of registered reference images, using recent retrieval techniques. This gives us a first estimate of the camera pose. To refine this estimate, like previous approaches, we match 2D points across the query image and the retrieved reference image. This step, however, is prone to fail as it is still very difficult to detect and match sparse feature points across images captured in potentially very different conditions. Our key contribution is to show that we need to extract sparse feature points only in the retrieved reference image: We then search for the corresponding 2D locations in the query image exhaustively. This search can be performed efficiently using convolutional operations, and robustly by using hypercolumn descriptors …",
author = "Hugo Germain and Guillaume Bourmaud and Vincent Lepetit",
year = "2019",
language = "English",
pages = "513--523",
booktitle = "2019 International Conference on 3D Vision (3DV)",

}

TY - GEN

T1 - Sparse-to-Dense Hypercolumn Matching for Long-Term Visual Localization

AU - Germain, Hugo

AU - Bourmaud, Guillaume

AU - Lepetit, Vincent

PY - 2019

Y1 - 2019

N2 - We propose a novel approach to feature point matching, suitable for robust and accurate outdoor visual localization in long-term scenarios. Given a query image, we first match it against a database of registered reference images, using recent retrieval techniques. This gives us a first estimate of the camera pose. To refine this estimate, like previous approaches, we match 2D points across the query image and the retrieved reference image. This step, however, is prone to fail as it is still very difficult to detect and match sparse feature points across images captured in potentially very different conditions. Our key contribution is to show that we need to extract sparse feature points only in the retrieved reference image: We then search for the corresponding 2D locations in the query image exhaustively. This search can be performed efficiently using convolutional operations, and robustly by using hypercolumn descriptors …

AB - We propose a novel approach to feature point matching, suitable for robust and accurate outdoor visual localization in long-term scenarios. Given a query image, we first match it against a database of registered reference images, using recent retrieval techniques. This gives us a first estimate of the camera pose. To refine this estimate, like previous approaches, we match 2D points across the query image and the retrieved reference image. This step, however, is prone to fail as it is still very difficult to detect and match sparse feature points across images captured in potentially very different conditions. Our key contribution is to show that we need to extract sparse feature points only in the retrieved reference image: We then search for the corresponding 2D locations in the query image exhaustively. This search can be performed efficiently using convolutional operations, and robustly by using hypercolumn descriptors …

M3 - Conference contribution

SP - 513

EP - 523

BT - 2019 International Conference on 3D Vision (3DV)

ER -