Pith. sign in

REVIEW 3 major objections 4 minor 36 references

To Glue or Not to Glue? Classical vs Learned Image Matching for Mobile Mapping Cameras to Textured Semantic 3D Building Models

T0 review · 3 major / 4 minor · reviewed 2026-08-07 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

Pith's one-line read Camera-to-model localization against CityGML LoD2 facade textures works with learned matchers and fails with classical ones.

desk verdict A useful, honest new benchmark for camera-to-CityGML-texture localization, but the zero-inlier headline for classical matchers is partly a 10px-RANSAC-threshold artifact against the paper's own documented 10-30px alignment noise. read the letter →

arxiv 2505.17973 v1 pith:WUDSCFOH submitted 2025-05-23 cs.CV cs.LG

Simone Gaisbauer , Prabin Gyawali , Qilin Zhang , Olaf Wysocki , Boris Jutzi This is my paper · ORCID
classification cs.CVcs.LG
keywords featurematchingvisuallocalizationCityGMLLoD2textured3Dbuildingmodelslearnedfeatureshandcraftedabsoluteposeestimationmobilemapping
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

This paper asks whether classical handcrafted feature matchers (SIFT, ORB, AKAZE) can still serve visual localization against textured semantic 3D building models, or whether learned matchers are required. It builds a pipeline that matches mobile mapping and drone images to orthorectified facade textures extracted from CityGML LoD2 models, converts the matched texture pixels to world coordinates via the model's texture-coordinate mapping, and estimates the camera pose with PnP plus RANSAC. On two custom datasets (622 car pairs and 18 UAV pairs), the learned pipelines, especially SuperPoint combined with LightGlue, produce usable pose estimates, whereas the classical methods essentially fail, with zero to 12 RANSAC inliers and AUC values near zero to 0.16. The paper positions this as evidence that model-based visual localization against city models should use learned matching, while cautioning that accuracy is coarse because of decimeter-level inconsistencies between model and survey data.

What carries the argument

The load-bearing object is the georeferenced texture face of the CityGML model: each planar polygonal face carries an image and an st-coordinate mapping (Equations 1-5) that turns a texture pixel into an East-North-Up world coordinate. This is what converts 2D image matches into 3D-to-2D correspondences for PnP, making absolute pose estimation possible without rendering views or building a 3D point cloud. The comparison then rides on the number and quality of matches that each matcher can deliver across the orthorectified-texture-to-perspective-image domain gap.

What would settle it

Re-run the same 622 car and 18 UAV pairs after co-registering the CityGML model to the camera trajectories, or after replacing the fixed 10 px RANSAC threshold with one derived from the manual-correspondence errors, and count RANSAC inliers and pose AUC for SIFT+FLANN. If the classical pipelines produce inlier counts and AUCs comparable to the learned ones, the central claim fails; if they remain near zero, the claim is supported.

Watch

Extended reading notes

Core claim

On the paper's terms, the discovery is that the domain gap between perspective mobile mapping images and orthorectified model textures is tractable for learnable feature matchers but essentially intractable for classical handcrafted pipelines. Using facade textures warped onto CityGML LoD2 faces, the authors derive world coordinates for matched keypoints and recover absolute camera poses with a PnP solver under RANSAC. On the car dataset, SuperPoint+LightGlue reaches a mean precision of 0.247 at a 30 px reprojection threshold and a mean inlier count of 12, while SIFT+FLANN, ORB+NN, and AKAZE+NN achieve essentially zero. The UAV dataset is harder, but the ordering is the same, with SuperPoint+LightGlue and SuperPoint+SuperGlue leading and classical methods at zero precision. The authors conclude that textured semantic models can support coarse or relative positioning, but that highly accurate absolute pose from such models requires resolving model-data inconsistencies first.

Load-bearing premise

The comparison assumes that the fixed 10-pixel RANSAC threshold treats all methods fairly, even though the paper itself shows the building model and the camera trajectories are misaligned by decimeters, producing reprojection errors of tens of pixels for hand-picked matches.

Editorial extensions

If this is right

  • If the central claim is correct, switching from classical to SuperPoint+LightGlue is the first-order change for camera-to-model localization against LoD2 textures; on the car dataset it is the configuration with the best pose accuracy.
  • SIFT, ORB, and AKAZE should not be expected to initialize pose estimation against orthorectified facade textures, because their zero-inlier failures indicate they cannot serve as a fallback.
  • The proposed pipeline can produce coarse absolute poses from textured semantic models without rendered views or explicit 3D point clouds, using only the model's texture coordinates.
  • LoFTR and DISK+LightGlue are viable but not better than the two-step extractor-plus-attention pipeline on this task.
  • Classical binary features remain faster, but on these data the speed advantage buys no usable pose estimate.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • Beyond the paper, the zero-inlier results for classical methods should not be read as a universal verdict on SIFT/ORB/AKAZE: the paper's own manual-correspondence check shows georeferencing misalignment alone can push reprojection errors to tens of pixels, so a co-registered version of the benchmark is a natural test of whether classical matchers would recover.
  • A direct extension the authors do not pursue is to feed the matched facade world coordinates into a landmark-based SLAM or visual odometry system; the framework already outputs exactly the 3D-to-2D correspondences such estimators consume.
  • Because the planar LoD2 facade geometry makes PnP sensitive to out-of-plane details such as roofs and sills, combining textures from several adjacent faces or refining the model should improve pose accuracy at least as much as swapping matchers.
  • The reported AUC ceiling of roughly 0.09-0.16 on the custom data suggests headroom for domain adaptation: fine-tuning a matcher on pairs of orthorectified textures and perspective facade images could be tested against these baselines.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

3 major / 4 minor

Summary. This paper benchmarks seven feature extraction and matching pipelines—SIFT+FLANN, ORB+NN, AKAZE+NN, SuperPoint+SuperGlue, SuperPoint+LightGlue, DISK+LightGlue, and LoFTR—for absolute camera pose estimation from mobile mapping images matched against orthorectified facade textures of textured CityGML LoD2 building models. The authors first validate implementations on HPatches and MegaDepth-1500, then introduce two custom TUM2TWIN-based datasets: 622 car-texture pairs and 18 UAV-texture pairs. Correspondences between texture pixels and world coordinates are derived from the CityGML st-coordinate and face geometry, and absolute poses are estimated with PnP-RANSAC using a fixed inlier threshold t=10 px. Metrics include mean precision, AUC of translation/rotation errors, and mean inlier counts. The central claim is that learnable matchers, especially SuperPoint+LightGlue, vastly outperform classical matchers, which reportedly produce zero RANSAC inliers on the custom datasets.

Significance. If the conclusion holds, the paper provides a useful, domain-specific comparison and a reusable dataset extension for model-based visual localization against semantic 3D city models. The authors contribute open code, release their evaluation framework, and extend TUM2TWIN with camera-to-model pairs, which is valuable for the photogrammetry and visual localization communities. The generic benchmark results are broadly consistent with the literature, lending some credibility to the implementation. However, the central custom-dataset claim rests on a fixed RANSAC threshold that appears to be smaller than the documented georeferencing misalignment, so the main quantitative conclusion is not yet established.

major comments (3)
  1. [§3.3.1, §4.1, §6, Tables 4–5] The fixed RANSAC threshold t=10 px is load-bearing for the paper's central claim, and the manuscript itself provides evidence that this threshold is below the alignment noise. Section 6 and Figure 9 report that decimeter-level inconsistencies between the CityGML model and the camera trajectories cause reprojection errors of tens of pixels even for manually picked correspondences. Under a 10 px RANSAC gate, a correct classical match whose projection error falls in the 10–30 px band is rejected by construction. This matters empirically: Table 4 shows SIFT+FLANN with mPrec@30px=0.019 on the car-texture set while mInl=0, so some SIFT matches are correct at the 30 px evaluation threshold but are discarded by the 10 px RANSAC threshold. The reported 'zero to 12 RANSAC-inliers' and the resulting 'vastly outperform' claim may therefore reflect threshold placement rather than matcher quality. A threshold sweep (e.g., 5, 10, 20, 30 px) or the use of a robust estimator such as MAGSAC is required before concluding that classical methods fail completely.
  2. [§5.2, Tables 4–5] The custom-dataset evaluation reports only point estimates—mPrec, AUC, and mInl—with no error bars, confidence intervals, or statistical tests. The 622-pair car-texture set and especially the 18-pair UAV-texture set are likely to have considerable pair-to-pair variability, and differences among learned methods such as SuperPoint+SuperGlue versus SuperPoint+LightGlue, or DISK+LightGlue versus LoFTR, are small relative to that expected variability. The authors themselves caution that the UAV results 'might not be as representative,' yet the abstract generalizes the conclusion across 'our challenging custom datasets.' The paper should report per-pair error distributions, bootstrap confidence intervals, or paired significance tests, and should temper claims such as 'the handcrafted methods failed completely' unless the statistical evidence supports them.
  3. [§5.1, Tables 2–3] The paper notes that its MegaDepth-1500 AUC values for LoFTR and for SuperPoint+LightGlue differ from published values, attributing this to the Kornia implementation and different resizing. Because the custom-dataset conclusions depend on the correctness of these matcher implementations, the authors should clarify in the text whether the observed differences are within expected implementation variation and, ideally, include the original reference implementations for at least one representative learned method to rule out implementation-specific degradation.
minor comments (4)
  1. [§3.2.1, Eq. (1)] In Equation (1), the second range is written as 'u ∈ [0, height]' but should read 'v ∈ [0, height]'.
  2. [§5.2, paragraph after Table 4] The text says 'SuperGlue+LightGlue performs best,' but the table identifies the best method as SuperPoint+LightGlue; please correct the name.
  3. [Tables 4–5] The table captions should define mInl and explicitly state the fixed RANSAC threshold t=10 px, since the inlier counts are a central result and the threshold is not obvious from the table alone.
  4. [Abstract] The phrase 'zero to 12 RANSAC-inliers and zero to 0.16 area under the curve' is unclear outside the full experimental context; consider rephrasing to state the metric ranges and the threshold setting more transparently.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the paper is an empirical benchmark with independent held-out evaluation; the RANSAC-threshold concern is a correctness risk, not a circularity.

full rationale

The paper's claims are empirical measurements, not derivations. Matching methods are either classical or pre-trained externally, and no parameter is fitted to the custom datasets to manufacture the result. The pose-estimation pipeline (texture pixel to st-coordinates to world coordinates, then PnP with RANSAC) is a standard geometric forward computation whose output is compared against RTK and Pix4D ground truth. The fixed RANSAC threshold of 10 px (Sections 3.3.1 and 4.1) is applied uniformly to all methods; although Section 6 documents decimeter-level misalignment causing tens-of-pixel reprojection errors even for manual correspondences, this is a threat to the validity of the comparison (possibly penalizing classical methods), not a circular reduction of the conclusion into its inputs. Self-citations to TUM2TWIN (Wysocki et al. 2025) and to semantic-model definitions (Wysocki et al. 2024) are data provenance and background, not load-bearing proof of the empirical outcome. The paper explicitly reports that generic-benchmark results (HPatches, MegaDepth) do not fully match the literature, showing independent measurement. Therefore no step qualifies as self-definitional, fitted-input-as-prediction, or imported uniqueness.

Assumptions & free parameters 3 free parameters · 4 assumptions · 0 invented entities

No new physical or algorithmic entities are introduced. The central comparison depends on geodetic alignment assumptions and on untested generalization of pretrained models; the RANSAC threshold is the main manually chosen numerical parameter.

free parameters (3)
  • RANSAC inlier threshold t = 10 px
    Set to a fixed value for all methods and chosen manually from the inconsistencies between model and camera data; no range was tested (Sections 3.3.1, 4.1). It directly determines which matches count as inliers and therefore shapes the zero-to-12 inlier results.
  • Image resizing for dense matchers = 1024 px on the larger dimension for LoFTR and DISK
    Resizing affects matching density and runtime for Megadepth-1500 and TUM images; chosen following common practice rather than optimized (Section 4.3).
  • Texture quality filtering thresholds = not reported
    Texture images below a certain size or with absent optical information are excluded, but the thresholds are not quantified, so the pair construction is not fully auditable (Section 3.2.1).
assumptions (4)
  • domain assumption CityGML LoD2 facade faces are planar and texture st-coordinates map linearly to world coordinates via Equations 1 to 5.
    The entire geo-referencing of matched pixels relies on this texture-to-face mapping, with no verification against independent control points.
  • domain assumption Pre-trained learned matchers generalize from projective training images to orthorectified facade textures.
    Authors state reliance on generalization because no training data exists for this domain (Section 6).
  • domain assumption Ground-truth poses and model coordinates are mutually consistent within the evaluation tolerance.
    Section 6 reports decimeter-level differences between datasets and tens of pixels of reprojection error even for manually picked matches, which weakens this premise.
  • domain assumption The UAV height offset of -45.66 m correctly accounts for the Bessel-to-WGS84 ellipsoid difference.
    Applied as a fixed datum correction to align UAV reconstruction heights with the CityGML model (Section 4.2.2).

how reviews work

0 comments
Cite this review

Pith. "Pith review of To Glue or Not to Glue? Classical vs Learned Image Matching for Mobile Mapping Cameras to Textured Semantic 3D Building Models." pith.science (2026). https://pith.science/paper/WUDSCFOH

@misc{pith2026250517973,
  author       = {Pith},
  title        = {Pith review of: To Glue or Not to Glue? Classical vs Learned Image Matching for Mobile Mapping Cameras to Textured Semantic 3D Building Models},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/WUDSCFOH}},
  note         = {Machine review of arXiv:2505.17973}
}
read the original abstract

Feature matching is a necessary step for many computer vision and photogrammetry applications such as image registration, structure-from-motion, and visual localization. Classical handcrafted methods such as SIFT feature detection and description combined with nearest neighbour matching and RANSAC outlier removal have been state-of-the-art for mobile mapping cameras. With recent advances in deep learning, learnable methods have been introduced and proven to have better robustness and performance under complex conditions. Despite their growing adoption, a comprehensive comparison between classical and learnable feature matching methods for the specific task of semantic 3D building camera-to-model matching is still missing. This submission systematically evaluates the effectiveness of different feature-matching techniques in visual localization using textured CityGML LoD2 models. We use standard benchmark datasets (HPatches, MegaDepth-1500) and custom datasets consisting of facade textures and corresponding camera images (terrestrial and drone). For the latter, we evaluate the achievable accuracy of the absolute pose estimated using a Perspective-n-Point (PnP) algorithm, with geometric ground truth derived from geo-referenced trajectory data. The results indicate that the learnable feature matching methods vastly outperform traditional approaches regarding accuracy and robustness on our challenging custom datasets with zero to 12 RANSAC-inliers and zero to 0.16 area under the curve. We believe that this work will foster the development of model-based visual localization methods. Link to the code: https://github.com/simBauer/To\_Glue\_or\_not\_to\_Glue

Figures

Figures reproduced from arXiv: 2505.17973 by the authors.

Figure 1
Figure 1. Classical (top) vs learnable (bottom) feature matching [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. Camera-to-textured-model image matching overview. [PITH_FULL_IMAGE:figures/full_fig_p002_2.png] view at source ↗
Figure 4
Figure 4. Sketch of the texturing, coordinate systems and [PITH_FULL_IMAGE:figures/full_fig_p003_4.png] view at source ↗
Figures from the paper (6 more)
Figure 3
Figure 3. Figure 3: Exemplary textured semantic 3D building model. [PITH_FULL_IMAGE:figures/full_fig_p003_3.png]
Figure 6
Figure 6. Figure 6: Example image of a textured 3D model. 3.3.1 Pose Estimation The pose estimation is done using Perspective-n-Point with RANSAC outlier removal. Instead of testing a range of values, we set the RANSAC threshold t to a fixed value. The parameter is chosen according to the…
Figure 5
Figure 5. Figure 5: All the camera images that contain one facade element [PITH_FULL_IMAGE:figures/full_fig_p004_5.png]
Figure 7
Figure 7. Figure 7: Example for learnables UAV-image-to-texture [PITH_FULL_IMAGE:figures/full_fig_p006_7.png]
Figure 8
Figure 8. Figure 8: Car and UAV point cloud and facade edges (pink [PITH_FULL_IMAGE:figures/full_fig_p007_8.png]
Figure 9
Figure 9. Figure 9: Error between manually picked points and projected, [PITH_FULL_IMAGE:figures/full_fig_p007_9.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

36 extracted references · 33 canonical work pages

  1. [1]

    write newline

    " write newline "" before.all 'output.state := FUNCTION list.label.output make.list.label ", " * write FUNCTION article output.bibitem list.label.output list.year output new.block title output new.block journal emphasize output format.volume.number output pages output new.block url output new.block note output fin.entry FUNCTION book output.bibitem list.l...

  2. [2]

    F., Solutions, T., 2011

    Alcantarilla, P. F., Solutions, T., 2011. Fast explicit diffusion for accelerated features in nonlinear scale spaces. IEEE Trans. Patt. Anal. Mach. Intell , 34(7), 1281--1298

    work page 2011

  3. [3]

    Uav laser scanning and photogrammetry of tum downtown campus

    Anders, K., Wang, J., Wysocki, O., Huang, X., Liu, S., 2025. Uav laser scanning and photogrammetry of tum downtown campus

    work page 2025

  4. [4]

    Three things everyone should know to improve object retrieval

    Arandjelovi \'c , R., Zisserman, A., 2012. Three things everyone should know to improve object retrieval. 2012 IEEE conference on computer vision and pattern recognition, IEEE, 2911--2918

    work page 2012

  5. [5]

    Hpatches: A benchmark and evaluation of handcrafted and learned local descriptors

    Balntas, V., Lenc, K., Vedaldi, A., Mikolajczyk, K., 2017. Hpatches: A benchmark and evaluation of handcrafted and learned local descriptors. Proceedings of the IEEE conference on computer vision and pattern recognition, 5173--5182

    work page 2017

  6. [6]

    Surf: Speeded up robust features

    Bay, H., Tuytelaars, T., Van Gool, L., 2006. Surf: Speeded up robust features. Computer Vision--ECCV 2006: 9th European Conference on Computer Vision, Graz, Austria, May 7-13, 2006. Proceedings, Part I 9, Springer, 404--417

    work page 2006

  7. [7]

    D., Wu, W., 2018

    Bian, J., Yang, R., Liu, Y., Zhang, L., Cheng, M., Reid, I. D., Wu, W., 2018. MatchBench: An Evaluation of Feature Matchers. CoRR , abs/1808.02267. http://arxiv.org/abs/1808.02267

    arXiv 2018

  8. [8]

    Analyzing the Impact of Semantic LoD3 Building Models on Image-based Vehicle Localization

    Bieringer, A., Wysocki, O., Tuttas, S., Hoegner, L., Holst, C., 2024. Analyzing the Impact of Semantic LoD3 Building Models on Image-based Vehicle Localization. ISPRS Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences , 10, 55--62

    work page 2024

Show all 36 references
  1. [9]

    Superpoint: Self-supervised interest point detection and description

    DeTone, D., Malisiewicz, T., Rabinovich, A., 2018. Superpoint: Self-supervised interest point detection and description. Proceedings of the IEEE conference on computer vision and pattern recognition workshops, 224--236

    2018

  2. [10]

    o ger, G., Kolbe, T. H., Nagel, C., H \

    Gr \"o ger, G., Kolbe, T. H., Nagel, C., H \"a fele, K.-H., 2012. Ogc city geography markup language (citygml) encoding standard

    2012

  3. [11]

    A survey of feature matching methods

    Huang, Q., Guo, X., Wang, Y., Sun, H., Yang, L., 2024. A survey of feature matching methods. IET Image Processing , 18(6), 1385--1410

    2024

  4. [12]

    Lfm-3d: Learnable feature matching across wide baselines using 3d signals

    Karpur, A., Perrotta, G., Martin-Brualla, R., Zhou, H., Araujo, A., 2024. Lfm-3d: Learnable feature matching across wide baselines using 3d signals. Proc. 3DV'24

    2024

  5. [13]

    H., Donaubauer, A., 2021

    Kolbe, T. H., Donaubauer, A., 2021. Semantic 3D city modeling and BIM . W. Shi, M. F. Goodchild, M. Batty, M.-P. Kwan, A. Zhang (eds), Urban Informatics, Springer Singapore, Singapore, 609--636

    2021

  6. [14]

    Grounding image matching in 3d with mast3r

    Leroy, V., Cabon, Y., Revaud, J., 2024. Grounding image matching in 3d with mast3r

    2024

  7. [15]

    A robust O (n) solution to the perspective-n-point problem

    Li, S., Xu, C., Xie, M., 2012. A robust O (n) solution to the perspective-n-point problem. IEEE transactions on pattern analysis and machine intelligence , 34(7), 1444--1450

    2012

  8. [16]

    Megadepth: Learning single-view depth prediction from internet photos

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

    2018

  9. [17]

    L., 2014

    Lin, T.-Y., Maire, M., Belongie, S., Hays, J., Perona, P., Ramanan, D., Doll \'a r, P., Zitnick, C. L., 2014. Microsoft coco: Common objects in context. Computer vision--ECCV 2014: 13th European conference, zurich, Switzerland, September 6-12, 2014, proceedings, part v 13, Spr...

    2014

  10. [18]

    Lightglue: Local feature matching at light speed

    Lindenberger, P., Sarlin, P.-E., Pollefeys, M., 2023. Lightglue: Local feature matching at light speed. Proceedings of the IEEE/CVF International Conference on Computer Vision, 17627--17638

    2023

  11. [19]

    S., Fanta-Jende, P., 2024

    Loeper, Y., Gerke, M., Alamouri, A., Kern, A., Bajauri, M. S., Fanta-Jende, P., 2024. Visual localization in urban environments employing 3D city models. The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences , 48, 311--318

    2024

  12. [20]

    G., 2004

    Lowe, D. G., 2004. Distinctive image features from scale-invariant keypoints. International journal of computer vision , 60, 91--110

    2004

  13. [21]

    S., Sattler, T., Pollefeys, M., Jutzi, B., 2019

    Mueller, M. S., Sattler, T., Pollefeys, M., Jutzi, B., 2019. Image-to-image translation for enhanced feature matching, image retrieval and visual localization. ISPRS annals , IV-2/W7, 111–119

    2019

  14. [22]

    G., 2009

    Muja, M., Lowe, D. G., 2009. Fast approximate nearest neighbors with automatic algorithm configuration. VISAPP (1) , 2(331-340), 2

    2009

  15. [23]

    Meshloc: Mesh-based visual localization

    Panek, V., Kukelova, Z., Sattler, T., 2022. Meshloc: Mesh-based visual localization. European Conference on Computer Vision, Springer, 589--609

    2022

  16. [24]

    Visual localization using imperfect 3d models from the internet

    Panek, V., Kukelova, Z., Sattler, T., 2023. Visual localization using imperfect 3d models from the internet. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 13175--13186

    2023

  17. [25]

    GlueStick: Robust Image Matching by Sticking Points and Lines Together

    Pautrat, R., Su \'a rez, I., Yu, Y., Pollefeys, M., Larsson, V., 2023. GlueStick: Robust Image Matching by Sticking Points and Lines Together . International Conference on Computer Vision (ICCV)

    2023

  18. [26]

    Kornia: an open source differentiable computer vision library for pytorch

    Riba, E., Mishkin, D., Ponsa, D., Rublee, E., Bradski, G., 2020. Kornia: an open source differentiable computer vision library for pytorch. Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision, 3674--3683

    2020

  19. [27]

    Orb: An efficient alternative to sift or surf

    Rublee, E., Rabaud, V., Konolige, K., Bradski, G., 2011. Orb: An efficient alternative to sift or surf. 2011 International conference on computer vision, Ieee, 2564--2571

    2011

  20. [28]

    Superglue: Learning feature matching with graph neural networks

    Sarlin, P.-E., DeTone, D., Malisiewicz, T., Rabinovich, A., 2020. Superglue: Learning feature matching with graph neural networks. Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, 4938--4947

    2020

  21. [29]

    et al., 2018

    Sattler, T., Maddern, W., Toft, C., Torii, A., Hammarstrand, L., Stenborg, E., Safari, D., Okutomi, M., Pollefeys, M., Sivic, J. et al., 2018. Benchmarking 6dof outdoor visual localization in changing conditions. Proceedings of the IEEE conference on computer vision and patter...

    2018

  22. [30]

    L., Frahm, J.-M., 2016

    Sch\" o nberger, J. L., Frahm, J.-M., 2016. Structure-from-motion revisited. Conference on Computer Vision and Pattern Recognition (CVPR)

    2016

  23. [31]

    Loftr: Detector-free local feature matching with transformers

    Sun, J., Shen, Z., Wang, Y., Bao, H., Zhou, X., 2021. Loftr: Detector-free local feature matching with transformers. Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, 8922--8931

    2021

  24. [32]

    DISK: Learning local features with policy gradient

    Tyszkiewicz, M., Fua, P., Trulls, E., 2020. DISK: Learning local features with policy gradient. Advances in Neural Information Processing Systems , 33, 14254--14265

    2020

  25. [33]

    Vggt: Visual geometry grounded transformer

    Wang, J., Chen, M., Karaev, N., Vedaldi, A., Rupprecht, C., Novotny, D., 2025. Vggt: Visual geometry grounded transformer

    2025

  26. [34]

    H., 2024

    Wysocki, O., Schwab, B., Beil, C., Holst, C., Kolbe, T. H., 2024. Reviewing Open Data Semantic 3D City Models to Develop Novel 3D Reconstruction Methods. The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences , 48, 493--500

    2024

  27. [35]

    K., Greza, M., Zhang, Q., Zhu, J., Froech, T., Heeramaglore, M., Hijazi, I., Kanna, K., Pechinger, M., Chen, Z., Sun, Y., Segura, A

    Wysocki, O., Schwab, B., Biswanath, M. K., Greza, M., Zhang, Q., Zhu, J., Froech, T., Heeramaglore, M., Hijazi, I., Kanna, K., Pechinger, M., Chen, Z., Sun, Y., Segura, A. R., Xu, Z., AbdelGafar, O., Mehranfar, M., Yeshwanth, C., Liu, Y.-C., Yazdi, H., Wang, J., Auer, S., Ande...

    2025 arXiv

  28. [36]

    Local feature matching using deep learning: A survey

    Xu, S., Chen, S., Xu, R., Wang, C., Lu, P., Guo, L., 2024. Local feature matching using deep learning: A survey. Information Fusion , 107, 102344

    2024

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.