DP-Loc: Visual Localization in 2D Maps Using an Embedded Depth Prior

Recent advancements in cost-effective image-based localization using 2D maps have garnered significant attention, inspired by humans’ ability to navigate with such maps. This study addresses the limitations of monocular vision-based systems, specifically inaccurate depth information and l...

Full description

Saved in:
Bibliographic Details
Main Authors: Kyoung Eun Kim, Joo Yong Sim
Format: Article
Language:English
Published: IEEE 2024-01-01
Series:IEEE Access
Subjects:
Online Access:https://ieeexplore.ieee.org/document/10772238/
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1846129710993506304
author Kyoung Eun Kim
Joo Yong Sim
author_facet Kyoung Eun Kim
Joo Yong Sim
author_sort Kyoung Eun Kim
collection DOAJ
description Recent advancements in cost-effective image-based localization using 2D maps have garnered significant attention, inspired by humans’ ability to navigate with such maps. This study addresses the limitations of monocular vision-based systems, specifically inaccurate depth information and loss of geometric details, which hinder precise localization. We propose a novel neural network framework that incorporates a pretrained metric depth estimation model, such as Zoedepth, to accurately measure absolute distances and enhance map matching between 2D maps and images. Our approach introduces two key modules: an Explicit Depth Prior Fusion (EDPF) module, which constructs a depth score volume using depth maps, and an Implicit Depth Prior Fusion (IDPF) module, which integrates depth and semantic features early through positional encoding. These modules enable a single-layer-scale classifier to learn essential features for effective localization. Notably, the IDPF model with positional encoding showed over 10% performance improvement on the Mapillary dataset compared to the baseline, underscoring the advantages of combining semantic and geometric information. The proposed DP-Loc approach provides a cost-efficient solution for visual localization by leveraging publicly accessible 2D maps and monocular image inputs, making it applicable to autonomous driving, robotics, and augmented reality.
format Article
id doaj-art-c8075dba132245c38682447a201577ce
institution Kabale University
issn 2169-3536
language English
publishDate 2024-01-01
publisher IEEE
record_format Article
series IEEE Access
spelling doaj-art-c8075dba132245c38682447a201577ce2024-12-10T00:01:49ZengIEEEIEEE Access2169-35362024-01-011218157018157810.1109/ACCESS.2024.351004610772238DP-Loc: Visual Localization in 2D Maps Using an Embedded Depth PriorKyoung Eun Kim0https://orcid.org/0009-0007-5570-5170Joo Yong Sim1https://orcid.org/0000-0003-3779-7589Department of Mechanical Systems Engineering, Sookmyung Women’s University, Seoul, Republic of KoreaDepartment of Mechanical Systems Engineering, Sookmyung Women’s University, Seoul, Republic of KoreaRecent advancements in cost-effective image-based localization using 2D maps have garnered significant attention, inspired by humans’ ability to navigate with such maps. This study addresses the limitations of monocular vision-based systems, specifically inaccurate depth information and loss of geometric details, which hinder precise localization. We propose a novel neural network framework that incorporates a pretrained metric depth estimation model, such as Zoedepth, to accurately measure absolute distances and enhance map matching between 2D maps and images. Our approach introduces two key modules: an Explicit Depth Prior Fusion (EDPF) module, which constructs a depth score volume using depth maps, and an Implicit Depth Prior Fusion (IDPF) module, which integrates depth and semantic features early through positional encoding. These modules enable a single-layer-scale classifier to learn essential features for effective localization. Notably, the IDPF model with positional encoding showed over 10% performance improvement on the Mapillary dataset compared to the baseline, underscoring the advantages of combining semantic and geometric information. The proposed DP-Loc approach provides a cost-efficient solution for visual localization by leveraging publicly accessible 2D maps and monocular image inputs, making it applicable to autonomous driving, robotics, and augmented reality.https://ieeexplore.ieee.org/document/10772238/Image-based map matchingvisual localizationmonocular depth estimationfeature fusion
spellingShingle Kyoung Eun Kim
Joo Yong Sim
DP-Loc: Visual Localization in 2D Maps Using an Embedded Depth Prior
IEEE Access
Image-based map matching
visual localization
monocular depth estimation
feature fusion
title DP-Loc: Visual Localization in 2D Maps Using an Embedded Depth Prior
title_full DP-Loc: Visual Localization in 2D Maps Using an Embedded Depth Prior
title_fullStr DP-Loc: Visual Localization in 2D Maps Using an Embedded Depth Prior
title_full_unstemmed DP-Loc: Visual Localization in 2D Maps Using an Embedded Depth Prior
title_short DP-Loc: Visual Localization in 2D Maps Using an Embedded Depth Prior
title_sort dp loc visual localization in 2d maps using an embedded depth prior
topic Image-based map matching
visual localization
monocular depth estimation
feature fusion
url https://ieeexplore.ieee.org/document/10772238/
work_keys_str_mv AT kyoungeunkim dplocvisuallocalizationin2dmapsusinganembeddeddepthprior
AT jooyongsim dplocvisuallocalizationin2dmapsusinganembeddeddepthprior