Pith. sign in

REVIEW 4 major objections 5 minor 2 cited by

SupeRANSAC: One RANSAC to Rule Them All

T0 review · 4 major / 5 minor · reviewed 2026-08-07 · deepseek-v4-flash

Pith's one-line read SupeRANSAC, a single unified RANSAC pipeline, is claimed to achieve the highest accuracy across homography, fundamental and essential matrix, rigid pose, and absolute pose estimation on 11 public large-scale datasets.

desk verdict A well-built unified RANSAC that likely works as advertised, but the test-set tuning leak and missing per-dataset numbers mean the headline gains need a clean-split rerun before I'd trust them. read the letter →

arxiv 2506.04803 v1 pith:MNFJHTMR submitted 2025-06-05 cs.CV

classification cs.CV
keywords RANSACrobustestimationhomographyfundamentalmatrixessentialabsoluteposerigidMAGSAC++
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 argues that the accuracy of RANSAC-style robust estimators is decided less by any single innovation such as sampling, scoring, or refinement, and more by how the full pipeline is assembled. It introduces SupeRANSAC, a single unified pipeline that combines PROSAC and P-NAPSAC sampling, MAGSAC++ scoring, GC-RANSAC local optimization, IRLS final optimization, and problem-specific minimal and nonminimal solvers. The central empirical claim is that this one configuration consistently reaches the highest accuracy on five geometric vision problems across 11 public large-scale datasets, improving on the best existing methods by, for example, 6 AUC points on average for fundamental matrix estimation. If correct, practitioners would not need to choose a different robust estimator per task, and the paper also implies that under-reported components such as local and final optimization matter more than the headline sampling or scoring method.

What carries the argument

The central object is the complete SupeRANSAC pipeline, whose load-bearing pieces are the scoring and optimization stages. The default scoring function is MAGSAC++, which marginalizes over a range of noise scales instead of committing to a single inlier-outlier threshold. Local optimization defaults to GC-RANSAC, which uses graph-cut to segment inliers from outliers followed by nested RANSAC sampling of size seven times the minimal sample, with a fallback to plain nested RANSAC when the correspondence count exceeds 2000. Final optimization is iteratively reweighted least squares with robust Cauchy weights, halving the threshold and re-evaluating consensus; for fundamental matrices, the nonminimal solver already integrates Levenberg-Marquardt refinement under the rank-2 parameterization, and the ablations show that adding a separate IRLS stage can slightly hurt. These stages carry the accuracy gains, as the ablation studies demonstrate for scoring, local optimization, and final optimization choices.

What would settle it

Re-run all comparisons with the 200 tuning pairs per dataset excluded from the test set and with parameters tuned on a disjoint hold-out; if SupeRANSAC's reported AUC margins over the next-best method shrink or reverse, the central accuracy claim would be refuted.

Watch

Extended reading notes

Core claim

The central claim is that SupeRANSAC, a unification of existing RANSAC improvements rather than a fundamentally new sampling or scoring principle, outperforms the state of the art on all five tested geometric estimation problems. The framework selects P-NAPSAC for spatially coherent problems such as homography, rigid pose, and absolute pose, and PROSAC for epipolar geometry to avoid degenerate localized samples. It applies explicit sample degeneracy checks such as twisted quadrilaterals and collinear 3D points, model checks such as determinant bounds and proper-rotation verification, MAGSAC++ as the default scoring function, preemptive verification with an optimistic score upper bound, GC-RANSAC local optimization that falls back to nested RANSAC above 2000 correspondences, and IRLS with Cauchy weights as final optimization. The paper reports the highest accuracy on fundamental and essential matrix estimation over six datasets totaling 39,592 image pairs, on homography over those datasets plus the HEB benchmark, on absolute pose on Aachen Day-Night and InLoc, and on rigid pose on 3DMatch and 3DLoMatch.

Load-bearing premise

The claim rests on the assumption that tuning a handful of parameters on 200 image pairs per dataset and then evaluating on the full test set, without removing those pairs, does not inflate SupeRANSAC's measured accuracy.

Editorial extensions

If this is right

  • One RANSAC configuration could replace problem-specific pipelines for homography, fundamental and essential matrix, rigid pose, and absolute pose estimation, yielding accuracy gains without task-specific code.
  • The combination of MAGSAC++ scoring, GC-RANSAC local optimization, and IRLS final optimization becomes a strong default recipe for future robust estimation libraries.
  • For fundamental matrix estimation, the built-in Levenberg-Marquardt refinement in the nonminimal solver is sufficient, so pipeline designers should not assume that adding more optimization stages always helps.
  • Preemptive verification using an optimistic score upper bound is broadly applicable to any scoring function with a per-point quality bound and does not degrade accuracy.
  • The reported gains imply that widely used existing implementations still leave significant accuracy on the table for common geometric estimation tasks.

Reading between the lines

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

  • If the reported accuracy margins survive a fully disjoint tuning split, they suggest that SupeRANSAC's advantage comes from assembling known components carefully rather than from a new algorithmic idea, meaning other frameworks could adopt the same choices and close the gap.
  • The especially strong nighttime localization result suggests that threshold-robust scoring such as MAGSAC++ becomes more valuable when feature matching quality degrades, a testable hypothesis across other challenging conditions.
  • The deliberate choice of PROSAC over P-NAPSAC for epipolar geometry encodes a prior that spatial locality causes degeneracy in relative pose; a systematic study of when locality helps versus hurts could generalize sampling selection across problems.
  • Because the paper intentionally excludes learned outlier pruning, evaluating SupeRANSAC on top of learned pruning methods is a natural extension that may yield further gains or show that the pipeline's consensus reasoning makes them redundant.
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

4 major / 5 minor

Summary. The paper introduces SupeRANSAC, a unified RANSAC-style robust estimation pipeline that assembles existing components—PROSAC/P-NAPSAC sampling, sample and model degeneracy checks, MAGSAC++ scoring, GC-RANSAC-style and nested-RANSAC local optimization, IRLS final optimization, and task-specific minimal/non-minimal solvers—into a single framework for homography, fundamental and essential matrix, absolute pose, and rigid 3D registration. The central claim is that this engineered system consistently outperforms existing public frameworks (OpenCV, PoseLib, pyCOLMAP, GC-RANSAC, MAGSAC++, VSAC) across 11 large-scale datasets and two feature types, with headline improvements such as 6 AUC points on average for fundamental matrix estimation. The evaluation includes accuracy-runtime trade-offs, threshold sensitivity curves, and component ablations. Code is publicly released.

Significance. If the empirical claims withstand scrutiny, the paper makes a useful practical contribution: it demonstrates that a carefully engineered, problem-specific RANSAC pipeline can match or beat specialized estimators across multiple geometric tasks, and it provides a detailed qualitative account of which components matter. The experimental breadth is genuinely large: 39,592 image pairs for epipolar geometry, two feature types, a dedicated homography benchmark, visual localization benchmarks, and 3D registration, with ablations for scoring, local optimization, and final optimization. The public code release is a further strength. However, the evaluation protocol has a load-bearing weakness: hyperparameters were tuned on subsets of the test data without removing those subsets, and no error bars or significance tests are reported. Until the evaluation is rerun on a clean split, the headline 'consistently highest accuracy' claim is not fully supported.

major comments (4)
  1. [IV.C, Tables I-III, Figs. 5-6] The evaluation protocol has test-set leakage. Section IV.C states that 200 image pairs per dataset were randomly selected from the test sets, used to tune parameters, and then 'these tuning pairs were not subsequently removed from the test set.' This is a direct contamination risk for the central quantitative claim. The concern is not merely hypothetical: several reported margins are small, e.g., Table III (SP+LG) gives SupeRANSAC AUC@10 of 0.41 versus 0.39 for GC-RANSAC, and Table I gives 0.59 versus 0.53 for the next best method. A small systematic upward bias on the 200 tuned pairs per dataset could shift the aggregated averages by exactly this magnitude. The assertion that overfitting to 200 pairs is 'highly unlikely' is unquantified. Moreover, the ablation studies and threshold-sensitivity curves in Figs. 5-6 are computed on the same 1,200-pair tuning set, so they demonstrate internal consistency but not generalization. Please rerun the main comparison on a disjoint evaluation split (or remove tuning pairs from the reported aggregates), and provide per-dataset breakdowns so the reader can assess the size of any leakage effect.
  2. [Tables I-VI, Fig. 3] No error bars, confidence intervals, or statistical significance tests are reported for any of the main comparisons. RANSAC-based estimators are stochastic, and single-run AUC differences of 0.02-0.03 (e.g., Table III, SP+LG row: SupeRANSAC 0.41 vs. GC-RANSAC 0.39; Table IV: mAA 0.51 vs. 0.44) may be consistent with run-to-run variation. Because the paper's central claim is 'consistently highest accuracy,' the authors should report mean and standard deviation over multiple random seeds, or bootstrap confidence intervals, for at least the headline comparisons in Tables I-III, and should state how many repeats were used. Without this, the claimed margins cannot be distinguished from noise.
  3. [Abstract, Section IV.D, Tables V-VI] The headline claim of 'consistently highest accuracy' is stronger than the data in several places. In Table V, on the Aachen Day subset SupeRANSAC achieves 80.7 at the (0.25m, 2°) threshold while LO-RANSAC [COLMAP] achieves 88.5, a clear regression on one of the two absolute-pose benchmarks. In Table VI, on 3DMatch SupeRANSAC's registration recall of 92.0 is below MAGSAC++'s 92.5, although its rotation/translation errors are better. The paper should either temper the claim to 'consistently competitive' or provide an aggregate summary measure across all tasks showing a net improvement, so that the one-RANSAC-for-all thesis is evaluated fairly.
  4. [IV.C] The description of baseline tuning is under-specified. The text says parameters were 'optimized on this set' for all estimators, but it does not report the parameter ranges searched, the number of configurations evaluated per method, or the criterion used to select the final value. Since SupeRANSAC and the baselines have different scoring functions and coherence weights, the fairness of the comparison depends on these details. Please provide the tuning protocol (grid/range, number of evaluations, selection rule) for each method and problem type.
minor comments (5)
  1. [Tables I, II, III, IV, VI] The abbreviations 'GC-RSC' and 'LO-RSC' are used for GC-RANSAC and LO-RANSAC without being defined in the tables; please harmonize with the names used in the text.
  2. [References] Reference [25] lists a placeholder URL (gts.sourceforge.net) alongside the PoseLib GitHub link; the entry should be corrected.
  3. [Header] The manuscript header states 'Manuscript received April 19, 2005,' which appears to be a typo for 2025.
  4. [III.G] The equation for the optimistic score reuses s(k) for both the partial score and the optimistic score; a distinct symbol such as \hat{s}(k) would avoid the confusing notation.
  5. [Fig. 5] The axis labels in Figure 5 are poorly formatted, with items such as 'MAGSAC ACRANSAC' and '2.08 × 100' requiring careful reading; please clean up the label spacing and scientific notation.

Circularity Check

0 steps flagged · score 2.0 of 10

No significant circularity: SupeRANSAC is an empirical integration of published RANSAC components; self-citations are not load-bearing in the derivation, though the tuning-on-test-set protocol is a validity caveat.

full rationale

SupeRANSAC does not derive predictions from first principles; it assembles published components (PROSAC, P-NAPSAC, MAGSAC++, GC-RANSAC, SP-RANSAC) into a unified pipeline. The central claim of consistently higher accuracy is an empirical comparison against external baselines (OpenCV, PoseLib, pyCOLMAP, etc.) on public benchmarks. Self-citations to MAGSAC++, GC-RANSAC, P-NAPSAC, VSAC, and SP-RANSAC refer to independently published methods with their own evaluations; none is invoked as an unverified uniqueness theorem or as the sole justification for excluding alternatives. The one notable caveat is Section IV.c: 200 image pairs per dataset were used to tune inlier thresholds and coherence weights, and these tuning pairs were not removed from the test set. This can bias the reported AUC margins, but it is a statistical leakage / experimental-protocol concern, not a circular derivation in which an output is equivalent to an input by construction. The paper's equations (e.g., the fundamental matrix parameterization in Eq. 1) are standard and do not embed the target result. Therefore no specific circular step is identified; the score of 2 reflects minor self-citation density and the tuning caveat, not load-bearing circularity.

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

The paper introduces no new physical or mathematical entities. Its central claim rests on standard robust estimation assumptions and several tuned parameters. The most significant assumption is that tuning on a subset of the test set does not bias results. The free parameters are all empirical constants, none of which are derived from first principles.

free parameters (4)
  • Inlier-outlier threshold tau = Not stated numerically; tuned per problem on 200-pair subsets (Section IV.c, Figs. 6a-c)
    The threshold is tuned on a small subset of each test dataset and then fixed across datasets. The central accuracy comparisons depend on this choice.
  • GC-RANSAC spatial coherence weight = Not stated
    Tuned for methods using GC-RANSAC local optimization, including SupeRANSAC's default LO. Mentioned in Section IV.c as a tuned parameter.
  • Correspondence count threshold for LO switch = 2000
    SupeRANSAC switches from GC-RANSAC to nested RANSAC when the number of correspondences exceeds 2000 (Section III.H). This is an empirically set design choice.
  • Termination confidence = 99.9%
    Standard RANSAC termination uses a user-defined confidence, set to 99.9% (Section III.J). This affects runtime but not final accuracy much.
assumptions (4)
  • domain assumption The RANSAC framework with minimal solvers is an appropriate estimator for the tested geometric problems.
    The entire paper assumes robust estimation via random sampling is the right approach, as opposed to learning-based or global methods. This is a standard assumption in the field.
  • domain assumption MAGSAC++ scoring provides a better model quality measure than inlier counting or MSAC across all tested problems.
    The choice of MAGSAC++ as the default scoring is justified by the ablation in Fig. 5, but the ablation is run on the same tuning set used for parameter selection, so the comparison is not fully independent.
  • domain assumption The evaluation metrics (AUC@10 degrees, mAA) from IMC and HEB benchmarks are meaningful measures of geometric accuracy.
    The paper uses these metrics as ground truth for correctness, which is standard but an assumption about what counts as good performance.
  • ad hoc to paper Sampling strategy choice per problem (P-NAPSAC for homography/rigid/absolute, PROSAC for epipolar) is beneficial.
    This design choice is stated in Section III.B based on geometric reasoning, but no ablation is shown for this specific decision, so it is an unverified assumption within the paper.

how reviews work

0 comments
Cite this review

Pith. "Pith review of SupeRANSAC: One RANSAC to Rule Them All." pith.science (2026). https://pith.science/paper/MNFJHTMR

@misc{pith2026250604803,
  author       = {Pith},
  title        = {Pith review of: SupeRANSAC: One RANSAC to Rule Them All},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/MNFJHTMR}},
  note         = {Machine review of arXiv:2506.04803}
}
read the original abstract

Robust estimation is a cornerstone in computer vision, particularly for tasks like Structure-from-Motion and Simultaneous Localization and Mapping. RANSAC and its variants are the gold standard for estimating geometric models (e.g., homographies, relative/absolute poses) from outlier-contaminated data. Despite RANSAC's apparent simplicity, achieving consistently high performance across different problems is challenging. While recent research often focuses on improving specific RANSAC components (e.g., sampling, scoring), overall performance is frequently more influenced by the "bells and whistles" (i.e., the implementation details and problem-specific optimizations) within a given library. Popular frameworks like OpenCV and PoseLib demonstrate varying performance, excelling in some tasks but lagging in others. We introduce SupeRANSAC, a novel unified RANSAC pipeline, and provide a detailed analysis of the techniques that make RANSAC effective for specific vision tasks, including homography, fundamental/essential matrix, and absolute/rigid pose estimation. SupeRANSAC is designed for consistent accuracy across these tasks, improving upon the best existing methods by, for example, 6 AUC points on average for fundamental matrix estimation. We demonstrate significant performance improvements over the state-of-the-art on multiple problems and datasets. Code: https://github.com/danini/superansac

Figures

Figures reproduced from arXiv: 2506.04803 by the authors.

Figure 1
Figure 1. Inliers found by SupeRANSAC for fundamental ma [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. Overview of the SupeRANSAC framework, where each processing stage is specialized for the given geometric estimation problem (e.g., homography or fundamental matrix estimation). All components are detailed in Sec. III. in each image: (1) no three points should be collinear, and (2) the quadrilateral formed by the four points (e.g., p1p2p3p4 in sequence) should not be ”twisted” or self-intersecting (i.e., it must be a… view at source ↗
Figure 3
Figure 3. Accuracy-runtime comparison of robust estimators for (a) fundamental matrix, (b) essential matrix, and (c) homography estimation. Each subplot displays the Area Under the recall Curve for pose error at a 10◦ threshold (AUC@10◦ ) against the average runtime in seconds. Results are averaged over six datasets: ScanNet1500 [54], [55], PhotoTourism [28], [56], LaMAR [57], 7Scenes [58], ETH3D [59], and KITTI [60], totalin… view at source ↗
Figures from the paper (3 more)
Figure 4
Figure 4. Figure 4: Mean Average Accuracy (mAA) scores versus runtime [PITH_FULL_IMAGE:figures/full_fig_p008_4.png]
Figure 5
Figure 5. Figure 5: Ablation studies of SupeRANSAC components. This figure illustrates the performance impact when varying key algorithmic choices. For each geometric problem shown – (a) Essential matrix, (b) Fundamental matrix, and (c) Homography (corresponding to image rows from top to …
Figure 6
Figure 6. Figure 6: Inlier-outlier threshold tuning for robust estimators across (a) fundamental matrix, (b) essential matrix, and (c) homography estimation. These plots illustrate Area Under the Curve (AUC@10◦ ) scores as a function of the inlier￾outlier threshold value (in pixels). Resu…

Discussion (0). Sign in to comment.

Forward citations

Cited by 2 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. SUFLECA: Scaling Up Feature Learning for CAD-to-image Alignment

    cs.CV 2026-07 conditional novelty 6.0 of 10

    SUFLECA achieves state-of-the-art single-image CAD-to-image alignment by scaling NOC-supervised feature learning to 674K images and adding geometrically consistent correspondence filtering.

  2. DS-SAC: Density Search for Sample Consensus

    cs.CV 2026-07 conditional novelty 6.0 of 10

    DS-SAC deterministically estimates robust geometric models via residual-density search and recursive partitioning, outperforming RANSAC variants in AUC and speed on large vision datasets.

Reference graph

Works this paper leans on

76 extracted references · 68 canonical work pages · cited by 2 Pith papers

  1. [1]

    From coarse to fine: Robust hierarchical localization at large scale,

    P.-E. Sarlin, C. Cadena, R. Siegwart, and M. Dymczyk, “From coarse to fine: Robust hierarchical localization at large scale,” in IEEE Conf. Comput. Vis. Pattern Recog., 2019, pp. 12 716–12 725

  2. [2]

    Structure-from-motion revisited,

    J. L. Schonberger and J.-M. Frahm, “Structure-from-motion revisited,” in IEEE Conf. Comput. Vis. Pattern Recog. , 2016, pp. 4104–4113

  3. [3]

    Global structure-from-motion revisited,

    L. Pan, D. Bar ´ath, M. Pollefeys, and J. L. Sch ¨onberger, “Global structure-from-motion revisited,” in European Conference on Computer Vision (ECCV), 2024

  4. [4]

    ORB-SLAM: a versatile and accurate monocular SLAM system,

    R. Mur-Artal, J. M. M. Montiel, and J. D. Tardos, “ORB-SLAM: a versatile and accurate monocular SLAM system,” IEEE Transactions on robotics, vol. 31, no. 5, pp. 1147–1163, 2015

  5. [5]

    RANSAC for robotic applications: A survey,

    J. M. Mart ´ınez-Otzeta, I. Rodr ´ıguez-Moreno, I. Mendialdua, and B. Sierra, “RANSAC for robotic applications: A survey,” Sensors, vol. 23, no. 1, p. 327, 2022. 14 2 4 6 8 10 Threshold [px] 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 AUC @ 10 Threshold vs Accuracy [SP+LG] SupeRANSAC RANSAC [OpenCV] RANSAC [skimage] GC-RANSAC [Author] GC-RANSAC [OpenCV] MAGSAC [A...

  6. [6]

    Epos: Estimating 6d pose of objects with symmetries,

    T. Hodan, D. Barath, and J. Matas, “Epos: Estimating 6d pose of objects with symmetries,” in IEEE Conf. Comput. Vis. Pattern Recog., 2020, pp. 11 703–11 712

  7. [7]

    Foundpose: Unseen object pose estimation with foundation features,

    E. P. ¨Ornek, Y . Labb´e, B. Tekin, L. Ma, C. Keskin, C. Forster, and T. Hodan, “Foundpose: Unseen object pose estimation with foundation features,” in Eur. Conf. Comput. Vis. Springer, 2025, pp. 163–182

  8. [8]

    Random sample consensus: a paradigm for model fitting with applications to image analysis and automated cartography,

    M. A. Fischler and R. C. Bolles, “Random sample consensus: a paradigm for model fitting with applications to image analysis and automated cartography,” Communications of the ACM, vol. 24, no. 6, pp. 381–395, 1981

Show all 76 references
  1. [9]

    RANSAC in 2020 tutorial,

    D. Barath, T.-J. Chin, O. Chum, D. Mishkin, R. Ranftl, and J. Matas, “RANSAC in 2020 tutorial,” in IEEE Conf. Comput. Vis. Pattern Recog. , 2020. [Online]. Available: http://cmp.felk.cvut.cz/ cvpr2020-ransac-tutorial/

  2. [10]

    Matching with prosac-progressive sample consensus,

    O. Chum and J. Matas, “Matching with prosac-progressive sample consensus,” in IEEE Conf. Comput. Vis. Pattern Recog. , vol. 1. IEEE, 2005, pp. 220–226

  3. [11]

    NAPSAC: High noise, high dimensional robust estimation-it’s in the bag,

    P. H. Torr, S. J. Nasuto, and J. M. Bishop, “NAPSAC: High noise, high dimensional robust estimation-it’s in the bag,” in Brit. Mach. Vis. Conf., vol. 2, 2002, p. 3

  4. [12]

    NeFSAC: Neurally filtered minimal samples,

    L. Cavalli, M. Pollefeys, and D. Barath, “NeFSAC: Neurally filtered minimal samples,” in Eur. Conf. Comput. Vis. Springer, 2022, pp. 351–366

  5. [13]

    Neural-guided RANSAC: Learning where to sample model hypotheses,

    E. Brachmann and C. Rother, “Neural-guided RANSAC: Learning where to sample model hypotheses,” inInt. Conf. Comput. Vis., 2019, pp. 4322– 4331

  6. [14]

    MLESAC: A new robust estimator with application to estimating image geometry,

    P. H. S. Torr and A. Zisserman, “MLESAC: A new robust estimator with application to estimating image geometry,” Comp. Vis. and Image Under., 2000

  7. [15]

    Marginalizing sample consensus,

    D. Barath, J. Noskova, and J. Matas, “Marginalizing sample consensus,” IEEE Trans. Pattern Anal. Mach. Intell., vol. 44, no. 11, pp. 8420–8432, 2021

  8. [16]

    MAGSAC++, a fast, reliable and accurate robust estimator,

    D. Barath, J. Noskova, M. Ivashechkin, and J. Matas, “MAGSAC++, a fast, reliable and accurate robust estimator,” in IEEE Conf. Comput. Vis. Pattern Recog., 2020, pp. 1304–1312

  9. [17]

    Learning to find good models in RANSAC,

    D. Barath, L. Cavalli, and M. Pollefeys, “Learning to find good models in RANSAC,” in IEEE Conf. Comput. Vis. Pattern Recog. , 2022, pp. 15 744–15 753

  10. [18]

    Locally optimized RANSAC,

    O. Chum, J. Matas, and J. Kittler, “Locally optimized RANSAC,” in Joint Pattern Recognition Symposium . Springer, 2003, pp. 236–243

  11. [19]

    Fixing the locally optimized RANSAC–full experimental evaluation,

    K. Lebeda, J. Matas, and O. Chum, “Fixing the locally optimized RANSAC–full experimental evaluation,” in British machine vision con- ference, vol. 2. Citeseer Princeton, NJ, USA, 2012

  12. [20]

    Two-view geometry estimation unaffected by a dominant plane,

    O. Chum, T. Werner, and J. Matas, “Two-view geometry estimation unaffected by a dominant plane,” in IEEE Conf. Comput. Vis. Pattern Recog., vol. 1. IEEE, 2005, pp. 772–779

  13. [21]

    DSAC - differentiable RANSAC for camera localiza- tion,

    E. Brachmann, A. Krull, S. Nowozin, J. Shotton, F. Michel, S. Gumhold, and C. Rother, “DSAC - differentiable RANSAC for camera localiza- tion,” in IEEE Conf. Comput. Vis. Pattern Recog. , 2017

  14. [22]

    Generalized differentiable RANSAC,

    T. Wei, Y . Patel, A. Shekhovtsov, J. Matas, and D. Barath, “Generalized differentiable RANSAC,” in Int. Conf. Comput. Vis. , 2023, pp. 17 649– 17 660

  15. [23]

    USAC: A universal framework for random sample consensus,

    R. Raguram, O. Chum, M. Pollefeys, J. Matas, and J.-M. Frahm, “USAC: A universal framework for random sample consensus,” IEEE Trans. Pattern Anal. Mach. Intell., vol. 35, no. 8, pp. 2022–2038, 2012

  16. [24]

    VSAC: Efficient and accurate estimator for h and f,

    M. Ivashechkin, D. Barath, and J. Matas, “VSAC: Efficient and accurate estimator for h and f,” in Int. Conf. Comput. Vis. , 2021, pp. 15 243– 15 252

  17. [25]

    PoseLib,

    V . Larsson, “PoseLib,” http://gts.sourceforge.net/https://github.com/ PoseLib/PoseLib, 2000–2004

  18. [26]

    Bradski, A

    G. Bradski, A. Kaehler et al., “Opencv,” Dr. Dobb’s journal of software tools, vol. 3, no. 2, 2000

  19. [27]

    Graph-cut RANSAC,

    D. Barath and J. Matas, “Graph-cut RANSAC,” in IEEE Conf. Comput. Vis. Pattern Recog., 2018, pp. 6733–6741

  20. [28]

    Photo tourism: exploring photo collections in 3d,

    N. Snavely, S. M. Seitz, and R. Szeliski, “Photo tourism: exploring photo collections in 3d,” in ACM siggraph 2006 papers , 2006, pp. 835–846

  21. [29]

    3dmatch: Learning local geometric descriptors from RGB-D recon- structions,

    A. Zeng, S. Song, M. Nießner, M. Fisher, J. Xiao, and T. Funkhouser, “3dmatch: Learning local geometric descriptors from RGB-D recon- structions,” in IEEE Conf. Comput. Vis. Pattern Recog., 2017, pp. 1802– 1811. 15

  22. [30]

    Superpoint: Self- supervised interest point detection and description,

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

  23. [31]

    Lightglue: Local feature matching at light speed,

    P. Lindenberger, P.-E. Sarlin, and M. Pollefeys, “Lightglue: Local feature matching at light speed,” in Int. Conf. Comput. Vis. , 2023, pp. 17 627– 17 638

  24. [32]

    Roma: Robust dense feature matching,

    J. Edstedt, Q. Sun, G. B ¨okman, M. Wadenb¨ack, and M. Felsberg, “Roma: Robust dense feature matching,” in IEEE Conf. Comput. Vis. Pattern Recog., 2024, pp. 19 790–19 800

  25. [33]

    Progressive napsac: sampling from gradually growing neighborhoods,

    D. Barath, M. Ivashechkin, and J. Matas, “Progressive napsac: sampling from gradually growing neighborhoods,” arXiv preprint arXiv:1906.02295, 2019

  26. [34]

    Finding geometric models by clustering in the consensus space,

    D. Barath, D. Rozumnyi, I. Eichhardt, L. Hajder, and J. Matas, “Finding geometric models by clustering in the consensus space,” in IEEE Conf. Comput. Vis. Pattern Recog., 2023, pp. 5414–5424

  27. [35]

    GroupSAC: Efficient consensus in the presence of groupings,

    K. Ni, H. Jin, and F. Dellaert, “GroupSAC: Efficient consensus in the presence of groupings,” in 2009 IEEE 12th International Conference on Computer Vision. IEEE, 2009, pp. 2193–2200

  28. [36]

    Adaptive sample consensus for efficient random optimization,

    L. Fan and T. Pylv ¨an¨ainen, “Adaptive sample consensus for efficient random optimization,” in ISVC. Springer, 2009, pp. 252–263

  29. [37]

    Bansac: A dynamic bayesian network for adaptive sample consensus,

    V . Piedade and P. Miraldo, “Bansac: A dynamic bayesian network for adaptive sample consensus,” in Int. Conf. Comput. Vis., 2023, pp. 3738– 3747

  30. [38]

    Adaptive reordering sampler with neurally guided magsac,

    T. Wei, J. Matas, and D. Barath, “Adaptive reordering sampler with neurally guided magsac,” in Int. Conf. Comput. Vis. , 2023, pp. 18 163– 18 173

  31. [39]

    Bayesian model estimation and selection for epipolar ge- ometry and generic manifold fitting,

    P. H. S. Torr, “Bayesian model estimation and selection for epipolar ge- ometry and generic manifold fitting,” International Journal of Computer Vision, vol. 50, pp. 35–61, 2002

  32. [40]

    MINPRAN: A new robust estimator for computer vision,

    C. V . Stewart, “MINPRAN: A new robust estimator for computer vision,” IEEE Trans. Pattern Anal. Mach. Intell. , vol. 17, no. 10, pp. 925–938, 1995

  33. [41]

    Automatic homographic reg- istration of a pair of images, with a contrario elimination of outliers,

    L. Moisan, P. Moulon, and P. Monasse, “Automatic homographic reg- istration of a pair of images, with a contrario elimination of outliers,” Image Processing On Line , vol. 2, pp. 56–73, 2012

  34. [42]

    Optimal randomized RANSAC,

    O. Chum and J. Matas, “Optimal randomized RANSAC,” IEEE Trans. Pattern Anal. Mach. Intell. , vol. 30, no. 8, pp. 1472–1482, 2008

  35. [43]

    Space-partitioning RANSAC,

    D. Barath and G. Valasek, “Space-partitioning RANSAC,” in Eur. Conf. Comput. Vis. Springer, 2022, pp. 721–737

  36. [44]

    RANSAC for (quasi-) degenerate data (QDEGSAC),

    J.-M. Frahm and M. Pollefeys, “RANSAC for (quasi-) degenerate data (QDEGSAC),” inIEEE Conf. Comput. Vis. Pattern Recog., vol. 1. IEEE, 2006, pp. 453–460

  37. [45]

    Gms: Grid-based motion statistics for fast, ultra-robust feature correspondence,

    J. Bian, W.-Y . Lin, Y . Matsushita, S.-K. Yeung, T.-D. Nguyen, and M.-M. Cheng, “Gms: Grid-based motion statistics for fast, ultra-robust feature correspondence,” in IEEE Conf. Comput. Vis. Pattern Recog. , 2017, pp. 4181–4190

  38. [46]

    Handcrafted outlier detection revisited,

    L. Cavalli, V . Larsson, M. R. Oswald, T. Sattler, and M. Pollefeys, “Handcrafted outlier detection revisited,” in ECCV. Springer, 2020, pp. 770–787

  39. [47]

    Learning to find good correspondences,

    K. M. Yi, E. Trulls, Y . Ono, V . Lepetit, M. Salzmann, and P. Fua, “Learning to find good correspondences,” in IEEE Conf. Comput. Vis. Pattern Recog., 2018, pp. 2666–2674

  40. [48]

    Pointnet: Deep learning on point sets for 3d classification and segmentation,

    C. R. Qi, H. Su, K. Mo, and L. J. Guibas, “Pointnet: Deep learning on point sets for 3d classification and segmentation,” in IEEE Conf. Comput. Vis. Pattern Recog., 2017, pp. 652–660

  41. [49]

    Acne: Atten- tive context normalization for robust permutation-equivariant learning,

    W. Sun, W. Jiang, E. Trulls, A. Tagliasacchi, and K. M. Yi, “Acne: Atten- tive context normalization for robust permutation-equivariant learning,” in IEEE Conf. Comput. Vis. Pattern Recog. , 2020, pp. 11 286–11 295

  42. [50]

    Learning two-view correspondences and geometry using order-aware network,

    J. Zhang, D. Sun, Z. Luo, A. Yao, L. Zhou, T. Shen, Y . Chen, L. Quan, and H. Liao, “Learning two-view correspondences and geometry using order-aware network,” in IEEE Conf. Comput. Vis. Pattern Recog., 2019, pp. 5845–5854

  43. [51]

    Progressive correspondence pruning by consensus learning,

    C. Zhao, Y . Ge, F. Zhu, R. Zhao, H. Li, and M. Salzmann, “Progressive correspondence pruning by consensus learning,” in Int. Conf. Comput. Vis., 2021, pp. 6464–6473

  44. [52]

    Ms2dg- net: Progressive correspondence learning via multiple sparse semantics dynamic graph,

    L. Dai, Y . Liu, J. Ma, L. Wei, T. Lai, C. Yang, and R. Chen, “Ms2dg- net: Progressive correspondence learning via multiple sparse semantics dynamic graph,” in IEEE Conf. Comput. Vis. Pattern Recog. , 2022, pp. 8973–8982

  45. [53]

    In defense of the eight-point algorithm,

    R. I. Hartley, “In defense of the eight-point algorithm,” IEEE Trans. Pattern Anal. Mach. Intell. , vol. 19, no. 6, pp. 580–593, 1997

  46. [54]

    Scannet: Richly-annotated 3d reconstructions of indoor scenes,

    A. Dai, A. X. Chang, M. Savva, M. Halber, T. Funkhouser, and M. Nießner, “Scannet: Richly-annotated 3d reconstructions of indoor scenes,” in IEEE Conf. Comput. Vis. Pattern Recog. , 2017, pp. 5828– 5839

  47. [55]

    Superglue: Learning feature matching with graph neural networks,

    P.-E. Sarlin, D. DeTone, T. Malisiewicz, and A. Rabinovich, “Superglue: Learning feature matching with graph neural networks,” in IEEE Conf. Comput. Vis. Pattern Recog., 2020, pp. 4938–4947

  48. [56]

    Image matching across wide baselines: From paper to practice,

    Y . Jin, D. Mishkin, A. Mishchuk, J. Matas, P. Fua, K. M. Yi, and E. Trulls, “Image matching across wide baselines: From paper to practice,” International Journal of Computer Vision , vol. 129, no. 2, pp. 517–547, 2021

  49. [57]

    Lamar: Benchmarking localization and mapping for augmented reality,

    P.-E. Sarlin, M. Dusmanu, J. L. Sch ¨onberger, P. Speciale, L. Gruber, V . Larsson, O. Miksik, and M. Pollefeys, “Lamar: Benchmarking localization and mapping for augmented reality,” in Eur. Conf. Comput. Vis. Springer, 2022, pp. 686–704

  50. [58]

    Real-time RGB-D camera relocalization,

    B. Glocker, S. Izadi, J. Shotton, and A. Criminisi, “Real-time RGB-D camera relocalization,” in IEEE Int. Symp. on Mixed and Augm. Reality . IEEE, 2013, pp. 173–179

  51. [59]

    Bad SLAM: Bundle adjusted direct RGB-D SLAM,

    T. Schops, T. Sattler, and M. Pollefeys, “Bad SLAM: Bundle adjusted direct RGB-D SLAM,” in IEEE Conf. Comput. Vis. Pattern Recog. , 2019, pp. 134–144

  52. [60]

    Vision meets robotics: The kitti dataset,

    A. Geiger, P. Lenz, C. Stiller, and R. Urtasun, “Vision meets robotics: The kitti dataset,” The International Journal of Robotics Research , vol. 32, no. 11, pp. 1231–1237, 2013

  53. [61]

    Hartley and A

    R. Hartley and A. Zisserman, Multiple view geometry in computer vision. Cambridge university press, 2003

  54. [62]

    An efficient solution to the five-point relative pose problem,

    D. Nist ´er, “An efficient solution to the five-point relative pose problem,” IEEE Trans. Pattern Anal. Mach. Intell. , vol. 26, no. 6, pp. 756–770, 2004

  55. [63]

    Sur deux problemes de reconstruction,

    M. Demazure, “Sur deux problemes de reconstruction,” Ph.D. disserta- tion, INRIA, 1988

  56. [64]

    J. C. Gower and G. B. Dijksterhuis, Procrustes problems. OUP Oxford, 2004, vol. 30

  57. [65]

    Lambda twist: An accurate fast robust perspective three point (P3P) solver,

    M. Persson and K. Nordberg, “Lambda twist: An accurate fast robust perspective three point (P3P) solver,” in Eur. Conf. Comput. Vis., 2018, pp. 318–332

  58. [66]

    H. O. Fattorini, The cauchy problem . Cambridge University Press, 1983, vol. 13517

  59. [67]

    A large- scale homography benchmark,

    D. Barath, D. Mishkin, M. Polic, W. F ¨orstner, and J. Matas, “A large- scale homography benchmark,” in IEEE Conf. Comput. Vis. Pattern Recog., 2023, pp. 21 360–21 370

  60. [68]

    A minimal solution for relative pose with unknown focal length,

    H. Stew ´enius, D. Nist ´er, F. Kahl, and F. Schaffalitzky, “A minimal solution for relative pose with unknown focal length,” Image and Vision Computing, vol. 26, no. 7, pp. 871–877, 2008

  61. [69]

    Nonlinear estimation of the fundamental matrix with minimal parameters,

    A. Bartoli and P. Sturm, “Nonlinear estimation of the fundamental matrix with minimal parameters,” IEEE Trans. Pattern Anal. Mach. Intell. , vol. 26, no. 3, pp. 426–432, 2004

  62. [70]

    EPnP: An accurate o (n) solution to the p n p problem,

    V . Lepetit, F. Moreno-Noguer, and P. Fua, “EPnP: An accurate o (n) solution to the p n p problem,” International journal of computer vision , vol. 81, pp. 155–166, 2009

  63. [71]

    Sift-the scale invariant feature transform,

    G. Lowe, “Sift-the scale invariant feature transform,” Int. J, vol. 2, no. 91-110, p. 2, 2004

  64. [72]

    Benchmarking 6dof outdoor visual localization in changing conditions,

    T. Sattler, W. Maddern, C. Toft, A. Torii, L. Hammarstrand, E. Stenborg, D. Safari, M. Okutomi, M. Pollefeys, J. Sivic et al. , “Benchmarking 6dof outdoor visual localization in changing conditions,” in IEEE Conf. Comput. Vis. Pattern Recog., 2018, pp. 8601–8610

  65. [73]

    Inloc: Indoor visual localization with dense matching and view synthesis,

    H. Taira, M. Okutomi, T. Sattler, M. Cimpoi, M. Pollefeys, J. Sivic, T. Pajdla, and A. Torii, “Inloc: Indoor visual localization with dense matching and view synthesis,” in IEEE Conf. Comput. Vis. Pattern Recog., 2018, pp. 7199–7209

  66. [74]

    Geotransformer: Fast and robust point cloud registration with geometric transformer,

    Z. Qin, H. Yu, C. Wang, Y . Guo, Y . Peng, S. Ilic, D. Hu, and K. Xu, “Geotransformer: Fast and robust point cloud registration with geometric transformer,” IEEE Trans. Pattern Anal. Mach. Intell. , 2023

  67. [75]

    Open3d: A modern library for 3d data processing,

    Q.-Y . Zhou, J. Park, and V . Koltun, “Open3d: A modern library for 3d data processing,” arXiv preprint arXiv:1801.09847 , 2018

  68. [76]

    Predator: Registration of 3d point clouds with low overlap,

    S. Huang, Z. Gojcic, M. Usvyatsov, A. Wieser, and K. Schindler, “Predator: Registration of 3d point clouds with low overlap,” in IEEE Conf. Comput. Vis. Pattern Recog. , 2021, pp. 4267–4276

Pith tools

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