pith. sign in

arxiv: 2502.09795 · v3 · submitted 2025-02-13 · 💻 cs.CV · cs.RO

Geometry-aided Vision-based Localization of Future Mars Helicopters in Challenging Illumination Conditions

Dario Pisanti , Robert Hewitt , Roland Brockers , Georgios Georgakis This is my paper

Pith reviewed 2026-05-23 03:01 UTC · model grok-4.3

classification 💻 cs.CV cs.RO
keywords map-based localizationMars helicopterimage registrationillumination robustnessdeep learninggeometry-aided matchingsimulation frameworkplanetary navigation
0
0 comments X

The pith

Geometry-aided deep learning registers Mars helicopter images to orbital maps despite large lighting and scale differences.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

Future Mars rotorcraft require map-based localization to correct visual odometry drift on long flights, yet traditional systems fail when onboard images differ sharply in illumination from reference orbital maps. The paper presents Geo-LoFTR, a deep learning registration model that incorporates geometric constraints, trained on large numbers of images synthesized from real orbital maps under varying sun angles. Comprehensive tests show the system achieves higher localization accuracy than prior methods across lighting and scale variations, and it maintains performance over an entire simulated Martian day as well as on real Mars photographs.

Core claim

The authors claim that adding geometric consistency checks to a transformer-based image matcher produces reliable registrations between live low-altitude images and an orbital reference map even under strong illumination mismatch and scale change, and that training exclusively on images rendered from real Martian orbital data enables successful transfer to actual Mars imagery.

What carries the argument

Geo-LoFTR, a geometry-aided deep learning model that augments a feature transformer with explicit geometric consistency to register onboard images against an orbital reference map.

If this is right

  • Mars helicopters can maintain accurate position estimates across a full Martian day instead of being limited to narrow lighting windows.
  • Cumulative drift from visual odometry is reduced more reliably during extended flights under changing sun angles.
  • The same orbital-map simulation pipeline supports training for missions on other bodies that possess orbital imagery.
  • Localization accuracy holds when the vehicle changes altitude and therefore image scale relative to the reference map.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The method could apply to Earth UAVs that must localize in shadowed or twilight conditions using satellite maps.
  • Fusion with inertial measurements might further lower error when illumination variation is extreme.
  • Direct evaluation on actual Ingenuity flight images would reveal whether the sim-to-real performance gap is smaller than the simulation tests suggest.

Load-bearing premise

The simulation framework that renders images from real orbital maps produces data realistic enough for the trained model to generalize to real low-altitude Mars photographs taken under varying illumination.

What would settle it

High localization error on real Mars helicopter images captured under strong illumination mismatch from the reference map, while error remains low on the simulated test set, would disprove the generalization claim.

Figures

Figures reproduced from arXiv: 2502.09795 by Dario Pisanti, Georgios Georgakis, Robert Hewitt, Roland Brockers.

Figure 1
Figure 1. Figure 1: Given an ortho-projected map of the terrain and a simulated onboard image we aim to estimate the pose of a rotocraft operating on Mars. Assuming [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Architecture of Geo-LoFTR that uses as inputs the query image [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: View of the Jezero Crater’s DTM in MARTIAN. [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Gray scale images of a map tile from the Jezero crater site, rendered [PITH_FULL_IMAGE:figures/full_fig_p005_4.png] view at source ↗
Figure 3
Figure 3. Figure 3: The lighting computations are performed using the [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗
Figure 5
Figure 5. Figure 5: Tiles from orthographic maps at sun (AZ=0°, EL=5°) ( [PITH_FULL_IMAGE:figures/full_fig_p006_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: Cumulative distributions of the localization error of simulated Mars [PITH_FULL_IMAGE:figures/full_fig_p007_6.png] view at source ↗
Figure 8
Figure 8. Figure 8: Geo-LoFTR, Pre-trained LoFTR and SIFT matched keypoints displayed for a sample query image ( [PITH_FULL_IMAGE:figures/full_fig_p008_8.png] view at source ↗
Figure 9
Figure 9. Figure 9: Cumulative distributions of the localization error of simulated Mars [PITH_FULL_IMAGE:figures/full_fig_p008_9.png] view at source ↗
Figure 12
Figure 12. Figure 12: Sample nadir-pointing observations rendered at different Local Mean [PITH_FULL_IMAGE:figures/full_fig_p009_12.png] view at source ↗
Figure 13
Figure 13. Figure 13: Localization accuracy (@1m) as a function of Local Mean Solar [PITH_FULL_IMAGE:figures/full_fig_p009_13.png] view at source ↗
read the original abstract

Planetary exploration using aerial assets has the potential for unprecedented scientific discoveries on Mars. While NASA's Mars helicopter Ingenuity proved flight in Martian atmosphere is possible, future Mars rotorcraft will require advanced navigation capabilities for long-range flights. One such critical capability is Map-based Localization (MbL) which registers an onboard image to a reference map during flight to mitigate cumulative drift from visual odometry. However, significant illumination differences between rotorcraft observations and a reference map prove challenging for traditional MbL systems, restricting the operational window of the vehicle. In this work, we investigate a new MbL system and propose Geo-LoFTR, a geometry-aided deep learning model for image registration that is more robust under large illumination differences than prior models. The system is supported by a custom simulation framework that uses real orbital maps to produce large amounts of realistic images of the Martian terrain. Comprehensive evaluations show that our proposed system outperforms prior MbL efforts in terms of localization accuracy under significant lighting and scale variations. Furthermore, we demonstrate the validity of our approach across a simulated Martian day and on real Mars imagery. Code and datasets are available at: https://dpisanti.github.io/geo-loftr/.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 2 minor

Summary. The paper introduces Geo-LoFTR, a geometry-aided deep-learning variant of LoFTR for map-based localization (MbL) of future Mars rotorcraft. It augments a custom simulation pipeline that renders images from real orbital maps to train the model for robustness to large illumination and scale changes, then reports that the system outperforms prior MbL methods on both a simulated Martian day and real Mars imagery.

Significance. If the reported accuracy gains on real imagery are reproducible and not artifacts of domain mismatch, the approach could meaningfully extend the operational envelope of Mars helicopters by relaxing illumination constraints on MbL. The release of code and datasets is a positive contribution for reproducibility.

major comments (2)
  1. [Abstract, §4] Abstract and §4 (evaluation): the headline claim that the system 'outperforms prior MbL efforts in terms of localization accuracy' is presented without any quantitative metrics, error bars, baseline implementations, or evaluation protocol. This absence makes it impossible to assess whether the central claim is supported by the data.
  2. [§3.2, §5] §3.2 (simulation framework) and §5 (real-image results): the validity of the real-Mars-imagery results rests on the untested assumption that the custom renderer reproduces the photometric and geometric statistics of actual rotorcraft observations sufficiently well for the geometry-aided LoFTR to retain its reported gains. No sim-to-real validation (e.g., feature-distribution divergence, failure-mode analysis, or cross-domain ablation) is described.
minor comments (2)
  1. [Abstract] The abstract states that 'comprehensive evaluations' were performed; the main text should include a dedicated evaluation-protocol subsection that specifies the exact metrics, number of test cases, and comparison methods.
  2. [§3.1] Notation for the geometry-aided components (e.g., how the geometric prior is injected into LoFTR) should be defined explicitly with equations rather than prose descriptions.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive feedback on our manuscript. We address each major comment below and outline the revisions we will make to improve clarity and rigor.

read point-by-point responses

  1. Referee: [Abstract, §4] Abstract and §4 (evaluation): the headline claim that the system 'outperforms prior MbL efforts in terms of localization accuracy' is presented without any quantitative metrics, error bars, baseline implementations, or evaluation protocol. This absence makes it impossible to assess whether the central claim is supported by the data.

    Authors: We agree that the abstract would be strengthened by including key quantitative results. Section 4 already contains the full evaluation, including localization error metrics (mean/median translation and rotation errors), success rates under illumination and scale variations, error bars from repeated trials, and a protocol that compares against reimplemented baselines on both simulated and real data. To make the central claim immediately verifiable, we will revise the abstract to report the primary accuracy gains (e.g., percentage reduction in median error relative to the strongest baseline). revision: yes

  2. Referee: [§3.2, §5] §3.2 (simulation framework) and §5 (real-image results): the validity of the real-Mars-imagery results rests on the untested assumption that the custom renderer reproduces the photometric and geometric statistics of actual rotorcraft observations sufficiently well for the geometry-aided LoFTR to retain its reported gains. No sim-to-real validation (e.g., feature-distribution divergence, failure-mode analysis, or cross-domain ablation) is described.

    Authors: The simulation pipeline renders from real orbital maps to generate training data, while §5 reports direct evaluation on held-out real Mars imagery. We acknowledge that the manuscript does not currently include explicit sim-to-real diagnostics such as feature-distribution divergence or cross-domain ablations. We will add these analyses in the revision, including quantitative comparisons of feature statistics between simulated and real images and an ablation showing performance when the model is tested on real data after training on simulation. revision: yes

Circularity Check

0 steps flagged

No circularity in derivation or evaluation chain

full rationale

The paper presents an empirical MbL system (Geo-LoFTR) trained on images from a custom simulator using real orbital maps, then evaluated for accuracy on simulated Martian day sequences and real Mars imagery. No equations, parameter fits, or derivations are described that reduce a claimed output to an input by construction. Performance claims rest on standard train/test splits and comparative metrics rather than self-definitional loops or load-bearing self-citations. The sim-to-real transfer is an empirical assumption subject to correctness scrutiny but does not constitute circularity under the specified patterns.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

Abstract-only review supplies no information on model hyperparameters, training losses, or background assumptions; ledger left empty pending full text.

pith-pipeline@v0.9.0 · 5744 in / 1060 out tokens · 25853 ms · 2026-05-23T03:01:54.440740+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

44 extracted references · 44 canonical work pages · 1 internal anchor

  1. [1]

    Multi-modal image registration for localization in titan’s atmosphere

    Adnan Ansar and Larry Matthies. Multi-modal image registration for localization in titan’s atmosphere. In2009 IEEE/RSJ International Conference on Intelligent Robots and Systems , pages 3349–3354. IEEE, 2009

    work page 2009

  2. [2]

    Jonathan Bapst, T. J. Parker, J. Balaram, T. Tzanetos, L. H. Matthies, C. D. Edwards, A. Freeman, S. Withrow- Maser, W. Johnson, E. Amador-French, J. L. Bishop, I. J. Daubar, C. M. Dundas, A. A. Fraeman, C. W. Hamilton, C. Hardgrove, B. Horgan, C. W. Leung, Y . Lin, A. Mittelholz, and B. P. Weiss. Mars Science Helicopter: Compelling Science Enabled by an ...

    work page 2021

  3. [3]

    Surf: Speeded up robust features

    Herbert Bay, Tinne Tuytelaars, and Luc Van Gool. Surf: Speeded up robust features. In Ale ˇs Leonardis, Horst Bischof, and Axel Pinz, editors, Computer Vision – ECCV 2006 , pages 404–417, Berlin, Heidelberg, 2006. Springer Berlin Heidelberg. ISBN 978-3-540-33833-8

    work page 2006

  4. [4]

    Golombek, Roland Brockers, Michael Mischna, and Martin R

    Roland Brockers, Pedro Proenc ¸a, Jeff Delaune, Jessica Todd, Larry Matthies, Theodore Tzanetos, and J. Bob Balaram. On-board absolute localization based on orbital imagery for a future mars science helicopter. In 2022 IEEE Aerospace Conference (AERO) , pages 1–11, 2022. doi: 10.1109/AERO53065.2022.9843673

    work page doi:10.1109/aero53065.2022.9843673 2022

  5. [5]

    Simultaneous orientation and scale estimator

    Yang Cheng and Adnan Ansar. Simultaneous orientation and scale estimator. 2024

    work page 2024

  6. [6]

    Blender hirise dtm importer, 2024

    PhaseIV contributors. Blender hirise dtm importer, 2024. URL https://github.com/phaseIV/ Blender-Hirise-DTM-Importer. Version 1.0

    work page 2024

  7. [7]

    A review on deep learning for uav absolute visual localization

    Andy Couturier and Moulay A Akhloufi. A review on deep learning for uav absolute visual localization. Drones, 8(11):622, 2024

    work page 2024

  8. [8]

    Vision-based uav self- positioning in low-altitude urban environments

    Ming Dai, Enhui Zheng, Zhenhua Feng, Lei Qi, Jiedong Zhuang, and Wankou Yang. Vision-based uav self- positioning in low-altitude urban environments. IEEE Transactions on Image Processing , 2023

    work page 2023

  9. [9]

    Superpoint: Self-supervised interest point de- tection and description

    Daniel DeTone, Tomasz Malisiewicz, and Andrew Rabi- novich. Superpoint: Self-supervised interest point de- tection and description. In Proceedings of the IEEE conference on computer vision and pattern recognition workshops, pages 224–236, 2018

    work page 2018

  10. [10]

    Astrovision: Towards autonomous feature detection and description for missions to small bodies using deep learning

    Travis Driver, Katherine A Skinner, Mehregan Dor, and Panagiotis Tsiotras. Astrovision: Towards autonomous feature detection and description for missions to small bodies using deep learning. Acta Astronautica, 210:393– 410, 2023

    work page 2023

  11. [11]

    Dkm: Dense kernelized feature matching for geometry estimation

    Johan Edstedt, Ioannis Athanasiadis, M ˚arten Wadenb¨ack, and Michael Felsberg. Dkm: Dense kernelized feature matching for geometry estimation. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 17765–17775, 2023

    work page 2023

  12. [12]

    Roma: Robust dense feature matching

    Johan Edstedt, Qiyu Sun, Georg B ¨okman, M ˚arten Wadenb¨ack, and Michael Felsberg. Roma: Robust dense feature matching. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 19790–19800, 2024

    work page 2024

  13. [13]

    Golombek, Roland Brockers, Michael Mischna, and Martin R

    H ˚avard Fjær Grip, Dylan Conway, Johnny Lam, Nathan Williams, Matthew P. Golombek, Roland Brockers, Michael Mischna, and Martin R. Cacan. Flying a heli- copter on mars: How ingenuity’s flights were planned, executed, and analyzed. In 2022 IEEE Aerospace Conference (AERO) , pages 1–17, 2022. doi: 10.1109/ AERO53065.2022.9843813

    work page arXiv 2022

  14. [14]

    Vision-based gnss-free localization for uavs in the wild

    Marius-Mihail Gurgu, Jorge Pe ˜na Queralta, and Tomi Westerlund. Vision-based gnss-free localization for uavs in the wild. In 2022 7th International Conference on Me- chanical Engineering and Robotics Research (ICMERR) , pages 7–12. IEEE, 2022

    work page 2022

  15. [15]

    Deep residual learning for image recognition

    Kaiming He, Xiangyu Zhang, Shaoqing Ren, and Jian Sun. Deep residual learning for image recognition. In Proceedings of the IEEE conference on computer vision and pattern recognition , pages 770–778, 2016

    work page 2016

  16. [16]

    Leveraging map retrieval and alignment for robust uav visual geo-localization

    Mengfan He, Jiacheng Liu, Pengfei Gu, and Ziyang Meng. Leveraging map retrieval and alignment for robust uav visual geo-localization. IEEE Transactions on Instrumentation and Measurement , 2024

    work page 2024

  17. [17]

    Johnson, Yang Cheng, Nikolas Trawny, James F

    Andrew E. Johnson, Yang Cheng, Nikolas Trawny, James F. Montgomery, Steven Schroeder, Johnny Chang, Daniel Clouse, Seth Aaron, and Swati Mohan. Im- plementation of a map relative localization system for planetary landing. Journal of Guidance, Control, and Dy- namics, 46(4):618–637, 2023. doi: 10.2514/1.G006780. URL https://doi.org/10.2514/1.G006780

    work page doi:10.2514/1.g006780 2023

  18. [18]

    Transformers are rnns: Fast au- toregressive transformers with linear attention

    Angelos Katharopoulos, Apoorv Vyas, Nikolaos Pappas, and Franc ¸ois Fleuret. Transformers are rnns: Fast au- toregressive transformers with linear attention. In Inter- national conference on machine learning , pages 5156–

    work page

  19. [19]

    Megadepth: Learning single-view depth prediction from internet photos

    Zhengqi Li and Noah Snavely. Megadepth: Learning single-view depth prediction from internet photos. In Proceedings of the IEEE conference on computer vision and pattern recognition , pages 2041–2050, 2018

    work page 2041

  20. [20]

    Lightglue: Local feature matching at light speed

    Philipp Lindenberger, Paul-Edouard Sarlin, and Marc Pollefeys. Lightglue: Local feature matching at light speed. In Proceedings of the IEEE/CVF International Conference on Computer Vision , pages 17627–17638, 2023

    work page 2023

  21. [21]

    David G. Lowe. Distinctive image features from scale-invariant keypoints. International Journal of Computer Vision , 60:91–110, 2004. URL https://api. semanticscholar.org/CorpusID:174065

    work page 2004

  22. [22]

    Jointloc: A real-time visual localization framework for planetary uavs based on joint relative and absolute pose estimation

    Xubo Luo, Xue Wan, Yixing Gao, Yaolin Tian, Wei Zhang, and Leizheng Shu. Jointloc: A real-time visual localization framework for planetary uavs based on joint relative and absolute pose estimation. arXiv preprint arXiv:2405.07429, 2024

    work page arXiv 2024

  23. [23]

    Aslfeat: Learning local features of accurate shape and localization

    Zixin Luo, Lei Zhou, Xuyang Bai, Hongkai Chen, Jiahui Zhang, Yao Yao, Shiwei Li, Tian Fang, and Long Quan. Aslfeat: Learning local features of accurate shape and localization. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , pages 6589– 6598, 2020

    work page 2020

  24. [24]

    McEwen, Eric M

    Alfred S. McEwen, Eric M. Eliason, James W. Bergstrom, Nathan T. Bridges, Candice J. Hansen, W. Alan Delamere, John A. Grant, Virginia C. Gulick, Kenneth E. Herkenhoff, Laszlo Keszthelyi, Randolph L. Kirk, Michael T. Mellon, Steven W. Squyres, Nicolas Thomas, and Catherine M. Weitz. Mars reconnaissance orbiter’s high resolution imaging science experiment ...

    work page doi:10.1029/2005je002605 2007

  25. [25]

    Mars24: Sunclock - a mars solar time and solar longitude cal- culator

    NASA Goddard Institute for Space Studies. Mars24: Sunclock - a mars solar time and solar longitude cal- culator. URL https://www.giss.nasa.gov/tools/mars24/. Version 8.3.1, released on 2023-05-18. Accessed: 2024- 10-28

    work page 2023

  26. [26]

    McClellan, J., Haghani, N., Winder, J., Huang, F., and Tokekar, P

    Jeremy Nash, Quintin Dwight, Lucas Saldyt, Haoda Wang, Steven Myint, Adnan Ansar, and Vandi Verma. Censible: A robust and practical global localization framework for planetary surface missions. In 2024 IEEE International Conference on Robotics and Automa- tion (ICRA) , pages 8642–8648, 2024. doi: 10.1109/ ICRA57147.2024.10611697

    work page arXiv 2024

  27. [27]

    DINOv2: Learning Robust Visual Features without Supervision

    Maxime Oquab, Timoth ´ee Darcet, Th ´eo Moutakanni, Huy V o, Marc Szafraniec, Vasil Khalidov, Pierre Fer- nandez, Daniel Haziza, Francisco Massa, Alaaeldin El- Nouby, et al. Dinov2: Learning robust visual features without supervision. arXiv preprint arXiv:2304.07193 , 2023

    work page internal anchor Pith review Pith/arXiv arXiv 2023

  28. [28]

    Rover relocalization for mars sample return by virtual template synthesis and matching

    Tu-Hoa Pham, William Seto, Shreyansh Daftry, Barry Ridge, Johanna Hansen, Tristan Thrush, Mark Van der Merwe, Gerard Maggiolino, Alexander Brinkman, John Mayo, et al. Rover relocalization for mars sample return by virtual template synthesis and matching. IEEE Robotics and Automation Letters , 6(2):4009–4016, 2021

    work page 2021

  29. [29]

    Illumination invariant image matching for lunar trn

    Noah Rothenberger, Georgios Georgakis, Yang Cheng, and Adnan Ansar. Illumination invariant image matching for lunar trn. In AIAA SCITECH 2025 F orum, page 2073, 2025

    work page 2025

  30. [30]

    Orb: An efficient alternative to sift or surf

    Ethan Rublee, Vincent Rabaud, Kurt Konolige, and Gary Bradski. Orb: An efficient alternative to sift or surf. In 2011 International Conference on Computer Vision, pages 2564–2571, 2011. doi: 10.1109/ICCV . 2011.6126544

    work page doi:10.1109/iccv 2011

  31. [31]

    Superglue: Learning feature matching with graph neural networks

    Paul-Edouard Sarlin, Daniel DeTone, Tomasz Mal- isiewicz, and Andrew Rabinovich. Superglue: Learning feature matching with graph neural networks. In Pro- ceedings of the IEEE/CVF conference on computer vision and pattern recognition , pages 4938–4947, 2020

    work page 2020

  32. [32]

    Loftr: Detector-free local feature matching with transformers

    Jiaming Sun, Zehong Shen, Yuang Wang, Hujun Bao, and Xiaowei Zhou. Loftr: Detector-free local feature matching with transformers. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 8922–8931, 2021

    work page 2021

  33. [33]

    Balaram, Shannah Withrow-Maser, Wayne Johnson, Larry Young, and Benjamin Pipenberg

    Theodore Tzanetos, Jonathan Bapst, Gerik Kubiak, Luis Phillipe Tosi, Sam Sirlin, Roland Brockers, Jeff Delaune, H ˚avard Fjær Grip, Larry Matthies, J. Balaram, Shannah Withrow-Maser, Wayne Johnson, Larry Young, and Benjamin Pipenberg. Future of mars rotorcraft - mars science helicopter. In 2022 IEEE Aerospace Conference (AERO), pages 1–16, 2022. doi: 10.1...

    work page doi:10.1109/aero53065 2022

  34. [34]

    Xue Wan, Jianguo Liu, Hongshi Yan, and Gareth L.K. Morgan. Illumination-invariant image matching for autonomous uav localisation based on optical sens- ing. ISPRS Journal of Photogrammetry and Re- mote Sensing , 119:198–213, 2016. ISSN 0924-

    work page 2016

  35. [35]

    doi: https://doi.org/10.1016/j.isprsjprs.2016.05

    work page doi:10.1016/j.isprsjprs.2016.05 2016

  36. [36]

    URL https://www.sciencedirect.com/science/article/ pii/S0924271616301113

    work page

  37. [37]

    Attention is all you need

    A Waswani, N Shazeer, N Parmar, J Uszkoreit, L Jones, A Gomez, L Kaiser, and I Polosukhin. Attention is all you need. In NIPS, 2017

    work page 2017

  38. [38]

    Vision-based uavs aerial image localiza- tion: A survey

    Yingxiao Xu, Long Pan, Chun Du, Jun Li, Ning Jing, and Jiangjiang Wu. Vision-based uavs aerial image localiza- tion: A survey. Proceedings of the 2nd ACM SIGSPATIAL International Workshop on AI for Geographic Knowledge Discovery, 2018. URL https://api.semanticscholar.org/ CorpusID:53428751

    work page 2018

  39. [39]

    Improving feature- based visual localization by geometry-aided matching

    Hailin Yu, Youji Feng, Weicai Ye, Mingxuan Jiang, Hujun Bao, and Guofeng Zhang. Improving feature- based visual localization by geometry-aided matching. arXiv preprint arXiv:2211.08712 , 2022

    work page arXiv 2022

  40. [40]

    Non-parametric local transforms for computing visual correspondence

    Ramin Zabih and John Woodfill. Non-parametric local transforms for computing visual correspondence. In Computer Vision—ECCV’94: Third European Confer- ence on Computer Vision Stockholm, Sweden, May 2–6 1994 Proceedings, V olume II 3, pages 151–158. Springer, 1994

    work page 1994

  41. [41]

    University- 1652: A multi-view multi-source benchmark for drone- based geo-localization

    Zhedong Zheng, Yunchao Wei, and Yi Yang. University- 1652: A multi-view multi-source benchmark for drone- based geo-localization. In Proceedings of the 28th ACM international conference on Multimedia , pages 1395– 1403, 2020

    work page 2020

  42. [42]

    Is geometry enough for matching in visual localization? In European Conference on Computer Vision, pages 407–425

    Qunjie Zhou, S ´ergio Agostinho, Aljo ˇsa Oˇsep, and Laura Leal-Taix´e. Is geometry enough for matching in visual localization? In European Conference on Computer Vision, pages 407–425. Springer, 2022

    work page 2022

  43. [43]

    Uav’s status is worth considering: A fusion representations matching method for geo- localization

    Runzhe Zhu, Mingze Yang, Ling Yin, Fei Wu, and Yuncheng Yang. Uav’s status is worth considering: A fusion representations matching method for geo- localization. Sensors, 23(2):720, 2023

    work page 2023

  44. [44]

    Sues-200: A multi-height multi- scene cross-view image benchmark across drone and satellite

    Runzhe Zhu, Ling Yin, Mingze Yang, Fei Wu, Yuncheng Yang, and Wenbo Hu. Sues-200: A multi-height multi- scene cross-view image benchmark across drone and satellite. IEEE Transactions on Circuits and Systems for Video Technology, 33(9):4825–4839, 2023

    work page 2023