Pith. sign in

REVIEW 3 major objections 5 minor 39 references

A Birotation Solution for Relative Pose Problems

T0 review · 3 major / 5 minor · reviewed 2026-08-16 · deepseek-v4-flash

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

Pith's one-line read The paper claims any relative pose can be recovered by minimizing three energy functions on the rotation manifold SO(3), using a birotation solution that avoids essential-matrix decomposition.

desk verdict A genuinely useful three-axis birotation extension with good engineering, but the paper's uniqueness story rests on a regularization term that the published update does not actually minimize. read the letter →

arxiv 2505.02025 v1 pith:RIM7XW6I submitted 2025-05-04 cs.CV

Hongbo Zhao , Ziwei Long , Mengtan Zhang , Hanli Wang , Qijun Chen , Rui Fan This is my paper · ORCID
classification cs.CV
keywords relativeposeestimationbirotationsolutionbasistransformationsRiemannianoptimizationSO(3)manifoldessentialmatrixvisualodometrystereocalibration
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

The paper sets out to show that relative camera pose, the rotation and translation between two views, can be estimated without ever forming or decomposing an essential matrix. It claims that any such pose can be captured by a birotation model: two rotation matrices that rotate the two camera coordinate systems until the remaining difference is a pure translation along one of three coordinate axes. Three geometric energy functions, one per axis, are minimized on the rotation manifold SO(3), and the lowest-energy solution yields the pose through closed-form recovery formulas. Because the model uses axial basis transformations rather than a single ideal stereo configuration, the paper argues it extends birotation ideas beyond stereo calibration to general indoor, outdoor, pure-rotation, and visual-odometry settings.

What carries the argument

The central object is the $i$-th birotation solution, a tuple $(R_1, R_2, s_i, l_i)$ satisfying $p_1^{C'} = p_2^{C'} + s_i l_i$ after rotating both views, where $l_1, l_2, l_3$ are the X-, Y-, and Z-axis unit vectors. The optimization machinery is Riemannian descent on SO(3): residuals are linearized using left-perturbation Jacobians, the increment $\Delta\Theta_i$ is solved in closed form by $\Delta\Theta_i = -(J_i \Lambda J_i^{\top} + \alpha I)^{-1} J_i \Lambda e_i$, and the rotations are updated by exponential maps. The key identity recovers the relative pose from whichever birotation solution has the minimal weighted residual.

What would settle it

Compute the three birotation solutions for a large set of image pairs, then restart each energy minimization from many random initial rotations; if the selected solution frequently changes or the relative pose derived from the minimal-energy basis disagrees with the ground truth while another solution agrees, the uniqueness and selection claims would be refuted. A simpler check is a synthetic scene whose true translation is far from all three axes: if at least one energy function systematically fails to converge to the ground-truth pose even from the RANSAC initialization, the coverage claim fails.

Watch

Extended reading notes

Core claim

The central claim is that the relative pose $(R,t)$ can be represented as $R = R_2^{\top} R_1$ and $t = -s_i R_2^{\top} l_i$, where $R_1$ and $R_2$ are rotations recovered by minimizing $E_i = e_i^{\top} \Lambda e_i + \alpha(\|\theta_{1,i}\|^2 + \|\theta_{2,i}\|^2)$ for $i = 1,2,3$. Here $e_i$ collects arctan-based residuals that measure how far the rotated correspondences are from aligning with a pure translation along the X-, Y-, or Z-axis, $\Lambda$ is an adaptive outlier mask, and $\alpha$ regularizes the two rotation vectors. The paper argues that each energy function has a well-defined local minimum on SO(3), that the three candidate minima bracketed by the basis transformations cover general relative poses, and that the residual-weighted selection rule identifies the correct basis. If true, this converts a five-degrees-of-freedom pose problem into a six-degrees-of-freedom birotation problem whose redundant gauge freedom the regularization resolves.

Load-bearing premise

The load-bearing premise is that, when started from the five-point RANSAC initialization, the regularized Riemannian descent converges to the correct minimum for at least one of the three energy functions and the selection rule recognizes it; the paper itself concedes in Section V that non-unique solutions remain a potential risk in practice.

Editorial extensions

If this is right

  • Relative pose can be recovered in closed form from a birotation solution via $R = R_2^{\top} R_1$ and $t = -s_i R_2^{\top} l_i$, so no essential-matrix decomposition is needed.
  • The method handles pure-rotation sequences, where essential-matrix methods degenerate, performing on par with homography-based baselines.
  • Arctan-based residuals eliminate pixel-location-dependent implicit weighting, improving accuracy for correspondences near image borders and in natural scenes.
  • With convergence parameters fixed, each iteration costs $O(N)$, and the method outperforms prior birotation and direct pose methods on indoor, outdoor, stereo-calibration, and visual-odometry benchmarks.
  • The three-energy setup is necessary: ablations show that optimizing only one energy function fails when the true motion is not aligned with that basis axis.

Reading between the lines

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

  • If the birotation parametrization is as general as reported, the same three-axis construction could be applied to pose problems beyond two cameras, such as multi-camera rigs or camera-to-robot hand-eye calibration, by treating each pair as a birotation subsystem.
  • The minimum residual that selects the basis could double as a confidence score: pose estimates with a clearly dominant basis are likely more reliable, which would be useful for outlier rejection or temporal smoothing in SLAM and visual odometry, though the paper does not make this claim.
  • Because the regularization resolves the redundant sixth degree of freedom, a natural test is whether warm-starting the optimization with the previous frame's solution in a video sequence improves both speed and temporal consistency; the paper does not report such an experiment.
  • A stronger conjecture is that the arctan-based residuals make the energy landscape benign enough that the RANSAC initialization could be replaced by a small number of random restarts; this is an inference, not a result in the paper.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

3 major / 5 minor

Summary. The manuscript introduces a 'birotation' parametrization for two-view relative pose estimation. Instead of estimating an essential matrix or a rotation/translation pair directly, the method represents the relative pose through two rotation matrices R1,i and R2,i that rotate the two camera coordinate systems into a configuration related by a pure translation along one of three coordinate axes. Three basis transformations (X-, Y-, and Z-axes) yield three geometric-metric energy functions, which are minimized on SO(3) with a Riemannian iteration; the solution with the lowest weighted residual is selected, and R and t are recovered via R = R2^T R1, t = -s_i R2^T l_i. The paper reports experiments on ScanNet, YFCC100M, KITTI, a pure-rotation dataset, stereo-calibration and visual-odometry tasks, and synthetic robustness tests.

Significance. If the method works as claimed, it provides a genuine alternative to essential-matrix-based two-view geometry that naturally handles pure rotation and near-degenerate configurations; the recovery formulas are derived from the birotation constraints in a self-contained way. The experimental coverage is broad, with dense and sparse correspondences, multiple baselines, ablations, initialization robustness, and synthetic noise studies. The paper's derivation of the residual Jacobians is carefully done, and the plan to release code and datasets is a positive sign. However, the paper's central optimization claim—that the L2 regularization term in the energy is minimized and guarantees uniqueness—is not supported by the stated update rule, and the performance claims need to be delimited because the YFCC100M results do not show universal superiority.

major comments (3)
  1. [Section III-D, Eqs. (36)-(48)] The iteration does not minimize the regularized energy E_i = e_i^T Λ e_i + α(||θ1,i||^2 + ||θ2,i||^2). The normal equation (47) is the damped Gauss-Newton step for the unregularized data term e_i^T Λ e_i only: it is obtained by setting the derivative of the first term of (42) to zero and adding αI as damping. The first-order condition for the stated objective contains an additional term α ∂R/∂Θ_i evaluated at the current iterate, so the published update satisfies J_i Λ e_i = 0 at a fixed point, not J_i Λ e_i + α ∂R/∂Θ_i = 0. As a consequence, α affects the iteration path but not the limit point, and the paper's assertion (Section III-D) that the regularization term 'ensure[s] the solution's uniqueness' is not established. The one-dimensional gauge ambiguity noted in Section III-C and the four-fold ambiguity discussed in Section V are therefore not resolved by the stated algorithm. The authors should either derive and use the correct regularized update (including the gradient of R with respect to the increment under the left-perturbation model) or reframe α as a damping parameter in a Levenberg-Marquardt iteration and substantially weaken the uniqueness claims.
  2. [Abstract and Section IV-D1, Table II] The claim of universal superior performance is not supported. On YFCC100M with dense correspondences, the proposed method is below Ling et al. [18] at @5 (45.89 vs 46.59) and @10 (63.45 vs 65.51); with sparse correspondences it is below [18] at all four thresholds (9.49 vs 10.00, 29.19 vs 31.13, 41.67 vs 45.13, 58.22 vs 63.53). The text correctly describes this as 'on par' (dense) and 'second-best' (sparse), but the abstract's 'superior performance' and the Introduction's 'greater accuracy over all other SoTA methods' overstate the results. Please revise the claims to name the regimes in which the method is actually state of the art.
  3. [Section V and Section VI] There is an internal tension between the Conclusion's claim that the method 'significantly overcame the limitations of existing birotation-based methods, particularly by ... alleviating the non-uniqueness in relative pose recovery' and the Section V concession that 'there remains a potential risk of yielding non-unique solutions in practical applications.' In light of the first major comment, the regularization term cannot be cited as the mechanism that removes this risk. The authors should either prove a basin-of-attraction or fixed-point uniqueness property for the actual update (47), or explicitly report the frequency of non-unique solutions in the experiments and remove the guarantee language.
minor comments (5)
  1. [Introduction, contributions] The third contribution says the algorithm 'simultaneously optimizes three energy functions', but Section III-D actually runs three independent optimizations in parallel and then selects among them; please rephrase to 'optimizes three energy functions in parallel'.
  2. [Section III-D, Eq. (32)] The essential-matrix expression for i=1 has a sign opposite to the derivation from [t]_x R; since the essential matrix is defined only up to scale this is harmless, but the sign should be made consistent or the scale ambiguity stated explicitly.
  3. [Section IV-A, paragraph 3] The citation [36] (Huber) does not appear to describe the 'upper quartile method' used to compute Λ; please provide a precise reference or a description of the outlier-removal procedure.
  4. [Abstract and Section I] The manuscript states that source code and datasets will be available upon publication, but no supplementary material or link is included; for review reproducibility, please provide an anonymized code appendix or a working link.
  5. [Fig. 5] Fig. 5 is very dense (11 sequences, 3 metrics, 8 methods); consider splitting it into sub-figures or providing the data in table form, since several 'Ours' values are difficult to read.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the birotation energy formulation and pose recovery are self-contained.

full rationale

The paper's core construction is self-contained: basis transformations are defined in Eq. (7), geometric metrics in Eqs. (24)-(26), residual vectors in Eqs. (33)-(34), energy functions in Eq. (36), and the final pose recovery in Eqs. (29)-(31) is an algebraic consequence of the birotation model. The residuals are not defined in terms of the output pose, and no constant is fitted to ground-truth poses in the derivation itself. The references to the authors' prior work [19] supply terminology and the left-perturbation Jacobian convention, but the Jacobians are given explicitly in Eqs. (44)-(46), so the derivation does not rely on an unverified self-citation. The empirical weights β_i in Eq. (49) are task-level selection priors rather than per-instance fitted parameters, and they only rank independently minimized candidate solutions rather than determining the pose directly. The possible discrepancy in Eq. (47), where the update omits the derivative of the L2 regularization term, is a correctness concern about which objective is actually minimized, not a circularity, because the energy functions and recovery formula are not defined in terms of the estimated pose. Therefore no load-bearing circular step is identifiable.

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

The method is largely self-contained given the calibrated-camera model. Hand-chosen items are the regularization weight, selection weights, convergence thresholds, and the upper-quartile outlier scheme; no invented entities are introduced. The main axiomatic input is the assumption of good initialization for the local manifold optimization, which the authors themselves flag as an open risk in Section V.

free parameters (3)
  • regularization weight alpha = 1e-3
    Used in Eqs. (36) and (47) to ensure uniqueness and reduce non-convexity; chosen by hand.
  • selection weights (beta1, beta2, beta3) = (0.25,1,1) for stereo, (1,1,0.25) for odometry, (1,1,1) otherwise
    Eq. (49); empirically set per task based on which basis transformation the ground-truth motion is expected to align with, leaking task prior into the result.
  • convergence thresholds = 10^-8 for value, 10^-6 for rate of change
    Termination criteria for the iterative optimization in Section III-D; chosen by hand.
assumptions (4)
  • domain assumption Cameras are calibrated with known intrinsic matrices K1 and K2
    The basis transformations and constraints in Eqs. (11)-(13) are formulated in normalized coordinates using K1 and K2; uncalibrated cases are outside scope.
  • ad hoc to paper The L2 regularization term with weight alpha ensures the local optimization converges to the intended basin
    Eq. (36) and Section III-D; the paper itself admits in Section V that uniqueness is not guaranteed and depends on initialization, so this is a pragmatic assumption rather than a proven property.
  • standard math Left perturbation model on SO(3) gives correct first-order gradients
    Eqs. (44)-(46) rely on the standard derivative of a rotated vector with respect to a left perturbation; this is standard manifold calculus.
  • standard math The epipolar constraint and essential matrix model are the accepted geometric model for calibrated two-view pose
    The whole formulation is built on the epipolar constraint in Eq. (2), which is standard in the field.

how reviews work

0 comments
Cite this review

Pith. "Pith review of A Birotation Solution for Relative Pose Problems." pith.science (2026). https://pith.science/paper/RIM7XW6I

@misc{pith2026250502025,
  author       = {Pith},
  title        = {Pith review of: A Birotation Solution for Relative Pose Problems},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/RIM7XW6I}},
  note         = {Machine review of arXiv:2505.02025}
}
abstract

Relative pose estimation, a fundamental computer vision problem, has been extensively studied for decades. Existing methods either estimate and decompose the essential matrix or directly estimate the rotation and translation to obtain the solution. In this article, we break the mold by tackling this traditional problem with a novel birotation solution. We first introduce three basis transformations, each associated with a geometric metric to quantify the distance between the relative pose to be estimated and its corresponding basis transformation. Three energy functions, designed based on these metrics, are then minimized on the Riemannian manifold $\mathrm{SO(3)}$ by iteratively updating the two rotation matrices. The two rotation matrices and the basis transformation corresponding to the minimum energy are ultimately utilized to recover the relative pose. Extensive quantitative and qualitative evaluations across diverse relative pose estimation tasks demonstrate the superior performance of our proposed birotation solution. Source code, demo video, and datasets will be available at \href{https://mias.group/birotation-solution}{mias.group/birotation-solution} upon publication.

Figures

Figures reproduced from arXiv: 2505.02025 by the authors.

Figure 1
Figure 1. (a) Definition of Φ = (ϕ1, ϕ2, ϕ3); (b) Comparison between ϵ1 and ϵ ′ 1 ; (c) Comparison between ϵ2 and ϵ ′ 2 ; (d) Comparison between ϵ3 and ϵ ′ 3 . ϵi , the i-th element of ϵ, is illustrated in [PITH_FULL_IMAGE:figures/full_fig_p004_1.png] view at source ↗
Figure 2
Figure 2. Visualizations of the normalized implicit weights [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. An overview of a single iteration in our proposed relative pose estimation algorithm. For each [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗
Figures from the paper (4 more)
Figure 4
Figure 4. Figure 4: Qualitative comparisons of relative pose algorithms on the stereo camera extrinsic calibration task: (a) Left images [PITH_FULL_IMAGE:figures/full_fig_p010_4.png]
Figure 5
Figure 5. Figure 5: Quantitative comparisons on the KITTI Odometry dataset, where “Seq [PITH_FULL_IMAGE:figures/full_fig_p011_5.png]
Figure 6
Figure 6. Figure 6: Comparisons of estimated trajectories on the KITTI Odometry dataset: (a) Estimated trajectories when correspondences [PITH_FULL_IMAGE:figures/full_fig_p012_6.png]
Figure 7
Figure 7. Figure 7: Average rotation and translation errors with respect to the correspondence noise levels and matching error rates. [PITH_FULL_IMAGE:figures/full_fig_p014_7.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

39 extracted references · 34 canonical work pages

  1. [12]

    An efficient solution to non-minimal case essential matrix estimation,

    J. Zhao, “An efficient solution to non-minimal case essential matrix estimation,” IEEE Transactions on Pattern Analysis and Machine Intel- ligence, vol. 44, no. 4, pp. 1777–1792, 2020

    work page 2020

  2. [18]

    High-precision online markerless stereo extrinsic calibration,

    Y . Ling and S. Shen, “High-precision online markerless stereo extrinsic calibration,” in IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) . IEEE, 2016, pp. 1771–1778

    work page 2016

  3. [1]

    Information-theoretic online multi-camera extrin- sic calibration,

    E. Dexheimer et al., “Information-theoretic online multi-camera extrin- sic calibration,” IEEE Robotics and Automation Letters , vol. 7, no. 2, pp. 4757–4764, 2022

    work page 2022

  4. [2]

    Continuous stereo self-calibration by camera parameter tracking,

    T. Dang et al., “Continuous stereo self-calibration by camera parameter tracking,” IEEE Transactions on Image Processing , vol. 18, no. 7, pp. 1536–1550, 2009

    work page 2009

  5. [3]

    Online continuous stereo extrinsic parameter estima- tion,

    P. Hansen et al., “Online continuous stereo extrinsic parameter estima- tion,” in Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR). IEEE, 2012, pp. 1059–1066

    work page 2012

  6. [5]

    Visual odometry and map correlation,

    A. Levin and R. Szeliski, “Visual odometry and map correlation,” in Proceedings of the IEEE Computer Society Conference on Com- puter Vision and Pattern Recognition (CVPR) . IEEE, 2004, DOI: 10.1109/CVPR.2004.1315088

    arXiv 2004

  7. [6]

    LEAP-VO: Long-term effective any point tracking for visual odometry,

    W. Chen et al. , “LEAP-VO: Long-term effective any point tracking for visual odometry,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) , 2024, pp. 19 844– 19 853

    work page 2024

  8. [7]

    In defense of the eight-point algorithm,

    R. I. Hartley, “In defense of the eight-point algorithm,” IEEE Transac- tions on Pattern Analysis and Machine Intelligence , vol. 19, no. 6, pp. 580–593, 1997

    work page 1997

Show all 39 references
  1. [8]

    An efficient solution to the five-point relative pose prob- lem,

    D. Nist ´er, “An efficient solution to the five-point relative pose prob- lem,” IEEE Transactions on Pattern Analysis and Machine Intelligence , vol. 26, no. 6, pp. 756–770, 2004

    2004

  2. [9]

    Optimal randomized RANSAC,

    O. Chum and J. Matas, “Optimal randomized RANSAC,” IEEE Trans- actions on Pattern Analysis and Machine Intelligence , vol. 30, no. 8, pp. 1472–1482, 2008

    2008

  3. [10]

    Generalized differentiable RANSAC,

    T. Wei et al., “Generalized differentiable RANSAC,” in Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV) , 2023, pp. 17 649–17 660

    2023

  4. [11]

    Deep keypoint-based camera pose estimation with geo- metric constraints,

    Y .-Y . Jauet al., “Deep keypoint-based camera pose estimation with geo- metric constraints,” in IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) . IEEE, 2020, pp. 4950–4957

    2020

  5. [13]

    MAGSAC: Marginalizing sample consensus,

    D. Barath et al. , “MAGSAC: Marginalizing sample consensus,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2019, pp. 10 197–10 205

    2019

  6. [14]

    Essential matrix estimation using Gauss-Newton iterations on a manifold,

    U. Helmke et al. , “Essential matrix estimation using Gauss-Newton iterations on a manifold,” International Journal of Computer Vision , vol. 74, pp. 117–136, 2007

    2007

  7. [15]

    Equivalent constraints for two-view geometry: Pose so- lution/pure rotation identification and 3D reconstruction,

    Q. Cai et al., “Equivalent constraints for two-view geometry: Pose so- lution/pure rotation identification and 3D reconstruction,” International Journal of Computer Vision , vol. 127, pp. 163–180, 2019

    2019

  8. [16]

    Revisiting the PnP problem: A fast, general and optimal solution,

    Y . Zheng et al. , “Revisiting the PnP problem: A fast, general and optimal solution,” in Proceedings of the IEEE International Conference on Computer Vision (ICCV) , 2013, pp. 2344–2351

    2013

  9. [17]

    EPnP: An accurate O(n) solution to the PnP problem,

    V . Lepetit et al., “EPnP: An accurate O(n) solution to the PnP problem,” International Journal of Computer Vision , vol. 81, pp. 155–166, 2009

    2009

  10. [19]

    Dive deeper into rectifying homography for stereo camera online self-calibration,

    H. Zhao et al. , “Dive deeper into rectifying homography for stereo camera online self-calibration,” in IEEE International Conference on Robotics and Automation (ICRA) . IEEE, 2024, pp. 14 479–14 485

    2024

  11. [20]

    Unified computation of strict maximum likelihood for geometric fitting,

    K. Kanatani and Y . Sugaya, “Unified computation of strict maximum likelihood for geometric fitting,” Journal of Mathematical Imaging and Vision, vol. 38, pp. 1–13, 2010. 16

    2010

  12. [21]

    Determining the epipolar geometry and its uncertainty: A review,

    Z. Zhang, “Determining the epipolar geometry and its uncertainty: A review,” International Journal of Computer Vision , vol. 27, pp. 161– 195, 1998

    1998

  13. [22]

    Optimization criteria and geometric algorithms for motion and structure estimation,

    Y . Ma et al., “Optimization criteria and geometric algorithms for motion and structure estimation,” International Journal of Computer Vision , vol. 44, pp. 219–249, 2001

    2001

  14. [23]

    The space of essential matrices as a Riemannian quotient manifold,

    R. Tron and K. Daniilidis, “The space of essential matrices as a Riemannian quotient manifold,” SIAM Journal on Imaging Sciences , vol. 10, no. 3, pp. 1416–1445, 2017

    2017

  15. [24]

    PoseNet: A convolutional network for real-time 6- DOF camera relocalization,

    A. Kendall et al. , “PoseNet: A convolutional network for real-time 6- DOF camera relocalization,” in Proceedings of the IEEE International Conference on Computer Vision (ICCV) , 2015, pp. 2938–2946

    2015

  16. [25]

    Unsupervised scale-consistent depth learning from video,

    J.-W. Bian et al. , “Unsupervised scale-consistent depth learning from video,” International Journal of Computer Vision , vol. 129, no. 9, pp. 2548–2564, 2021

    2021

  17. [26]

    SCIPaD: Incorporating spatial clues into unsupervised pose-depth joint learning,

    Y . Feng et al. , “SCIPaD: Incorporating spatial clues into unsupervised pose-depth joint learning,” IEEE Transactions on Intelligent Vehicles , 2024, DOI: 10.1109/TIV .2024.3460868

    2024

  18. [27]

    J. D. Dixon and B. Mortimer, Permutation groups. Springer Science & Business Media, 1996, vol. 163

    1996

  19. [28]

    LoFTR: Detector-free local feature matching with trans- formers,

    J. Sun et al., “LoFTR: Detector-free local feature matching with trans- formers,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) , 2021, pp. 8922–8931

    2021

  20. [29]

    MegaDepth: Learning single-view depth pre- diction from internet photos,

    Z. Li and N. Snavely, “MegaDepth: Learning single-view depth pre- diction from internet photos,” in Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR) , 2018, pp. 2041– 2050

    2018

  21. [30]

    Distinctive image features from scale-invariant keypoints,

    D. G. Lowe, “Distinctive image features from scale-invariant keypoints,” International Journal of Computer Vision , vol. 60, pp. 91–110, 2004

    2004

  22. [31]

    LightGlue: Local feature matching at light speed,

    P. Lindenberger et al. , “LightGlue: Local feature matching at light speed,” in Proceedings of the IEEE/CVF International Conference on Computer Vision, 2023, pp. 17 627–17 638

    2023

  23. [32]

    ScanNet: Richly-annotated 3D reconstructions of indoor scenes,

    A. Dai et al., “ScanNet: Richly-annotated 3D reconstructions of indoor scenes,” in Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR) , 2017, pp. 5828–5839

    2017

  24. [33]

    YFCC100M: The new data in multimedia research,

    B. Thomee et al., “YFCC100M: The new data in multimedia research,” Communications of the ACM , vol. 59, no. 2, pp. 64–73, 2016

    2016

  25. [34]

    Hartley and A

    R. Hartley and A. Zisserman, Multiple view geometry in computer vision. Cambridge University Press, 2003

    2003

  26. [35]

    Vision meets robotics: The KITTI dataset,

    A. Geiger et al. , “Vision meets robotics: The KITTI dataset,” The International Journal of Robotics Research , vol. 32, no. 11, pp. 1231– 1237, 2013

    2013

  27. [36]

    Robust estimation of a location parameter,

    P. J. Huber, “Robust estimation of a location parameter,” in Break- throughs in statistics: Methodology and distribution . Springer, 1992, pp. 492–518

    1992

  28. [37]

    SuperGlue: Learning feature matching with graph neural networks,

    P.-E. Sarlin et al. , “SuperGlue: Learning feature matching with graph neural networks,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) , 2020, pp. 4938– 4947

    2020

  29. [38]

    A benchmark for the evaluation of RGB-D SLAM systems,

    J. Sturm et al. , “A benchmark for the evaluation of RGB-D SLAM systems,” in 2012 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) . IEEE, 2012, pp. 573–580

    2012

  30. [39]

    Pyramid stereo matching network,

    J.-R. Chang and Y .-S. Chen, “Pyramid stereo matching network,” in Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2018, pp. 5410–5418

    2018

  31. [40]

    RAFT-Stereo: Multilevel recurrent field transforms for stereo matching,

    L. Lipson et al., “RAFT-Stereo: Multilevel recurrent field transforms for stereo matching,” in 2021 International Conference on 3D Vision (3DV). IEEE, 2021, pp. 218–227

    2021

Pith tools

Reviewed August 16, 2026 · model on record in the stance chip above.