Pith. sign in

REVIEW 4 major objections 5 minor 2 cited by

InstantSfM claims a fully GPU-native, PyTorch-compatible global Structure-from-Motion pipeline that runs up to 40x faster than COLMAP while keeping comparable accuracy.

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

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

T0 review · deepseek-v4-flash

2026-08-04 09:46 UTC pith:IBNTYOWB

load-bearing objection A credible GPU-native SfM efficiency win with an accuracy claim that overshoots on Tanks and Temples; worth refereeing with conditions. the 4 major comments →

arxiv 2510.13310 v3 pith:IBNTYOWB submitted 2025-10-15 cs.CV

InstantSfM: Towards GPU-Native SfM for the Deep Learning Era

Jiankun Zhong , Zitong Zhan , Quankai Gao , Ziyu Chen , Haozhe Lou , Jiageng Mao , Ulrich Neumann , Chen Wang
show 1 more author
Yue Wang
This is my paper
classification cs.CV
keywords Structure-from-MotionGPUBundle AdjustmentGlobal PositioningSparse OptimizationPyTorchDepth PriorLevenberg-Marquardt
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

The pith

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

InstantSfM proposes to rebuild the entire Structure-from-Motion pipeline on the GPU, using sparse-aware Levenberg-Marquardt optimization natively in PyTorch for both bundle adjustment and global positioning. The paper claims this removes the CPU bottleneck that limits classical SfM systems and allows thousands of images to be processed on a single GPU, with up to ~40x speedup over COLMAP. It also injects metric depth priors directly into the optimization via the Jacobian, resolving scale ambiguity so that cameras and point clouds come out at metric scale. If correct, this would make SfM a fast, flexible, differentiable component for deep learning and 3D reconstruction workflows.

Core claim

The central claim is that a complete global SfM system can be implemented entirely on the GPU and in PyTorch without sacrificing reconstruction quality. The authors build on sparse-aware bundle adjustment, extend it to global positioning, and add a depth-constrained Jacobian structure that lets metric depth priors resolve scale ambiguity within the optimization itself. They report up to ~40x speedup over COLMAP on large-scale scenes, comparable accuracy on several benchmarks, and robustness on indoor RGB-D scenes where COLMAP and GLOMAP often fail.

What carries the argument

The machinery is a sparse Jacobian representation combined with Levenberg-Marquardt steps whose expensive operations — JTJ, JT r, diagonal fetching, and the linear solve — are implemented with GPU sparse-matrix primitives. Depth priors enter by replacing the per-observation scale variable with 1/depth, so the Jacobian blocks gain depth-constraint terms. Explicit masking of under-constrained variables keeps the linear system nonsingular.

Load-bearing premise

The accuracy comparison assumes that Python-based view-graph calibration and OpenCV relative-pose estimation produce camera initializations as good as COLMAP/GLOMAP's C++ Ceres/PoseLib pipeline; the paper's own Tanks-and-Temples results show it sometimes does not.

What would settle it

Run InstantSfM and COLMAP on the same 300-frame Tanks-and-Temples scenes and compare relative-pose AUC; if InstantSfM's Playground AUC@5 stays below COLMAP's by more than a few points, the 'comparable accuracy' claim, as measured on that benchmark, fails even if the speedup holds.

Watch this falsifier. Get emailed when new claim-graph text bears on it.

If this is right

  • SfM can scale to thousands of images on a single GPU, where learning-based pipelines like VGGSfM run out of memory.
  • Learning-based feature descriptors can be used directly in Python, removing the C++ conversion step.
  • Supplying monocular or sensor depth gives metric-scale camera poses and point clouds, useful for robotics and simulation.
  • The sparse-aware optimization approach may accelerate other large-scale non-linear least squares problems beyond SfM.

Where Pith is reading between the lines

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

  • If the approach is adopted, SfM could become differentiable end-to-end, enabling joint training of feature extraction and geometric optimization.
  • The depth-prior trick suggests a general way to inject metric information into any scale-ambiguous geometric optimization, potentially generalizing to SLAM or multi-view stereo.
  • The reported accuracy gap on Tanks-and-Temples indicates initialization quality is the bottleneck; a drop-in replacement of PyCeres/OpenCV with C++-equivalent components would test that hypothesis.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

4 major / 5 minor

Summary. The paper presents InstantSfM, a GPU-native Structure-from-Motion pipeline implemented in PyTorch. It applies sparse-aware Levenberg-Marquardt optimization to both global positioning (GP) and bundle adjustment (BA), and optionally injects metric depth priors into the optimization to recover metric-scale reconstructions. The authors claim state-of-the-art efficiency—up to ~40x speedup over COLMAP on large scenes—and reconstruction accuracy comparable to established classical pipelines (COLMAP, GLOMAP) and learning-based methods (VGGSfM). Experiments span MipNeRF360, DTU, Tanks and Temples, ScanNet, ScanNet++, plus runtime benchmarks on BAL and 1DSfM. The efficiency results are strong, but the accuracy comparison is mixed: on Tanks and Temples the proposed method is consistently below COLMAP and GLOMAP on every scene and threshold, while on MipNeRF360 and DTU downstream rendering metrics are more favorable. The paper also introduces a depth-prior mechanism (Eq. 4) whose definition and Jacobian structure are under-specified.

Significance. If the efficiency results hold, the paper makes a practically important contribution: a sparse, GPU-native, PyTorch-compatible SfM pipeline that handles thousands of images on a single GPU, removing the CPU bottleneck of COLMAP/GLOMAP. The runtime comparisons in Tables VI and VII and Fig. 3 show order-of-magnitude speedups over COLMAP on large scenes, and the ScanNet experiments demonstrate robustness in cases where COLMAP and GLOMAP fail. The depth-prior ablation is a useful step toward metric-scale SfM. However, the central 'comparable accuracy' claim is not uniformly supported: Table III shows clear end-to-end accuracy deficits on Tanks and Temples, and the depth-prior formulation is not fully specified. The paper's value is therefore currently concentrated in the efficiency/robustness results rather than the accuracy-equivalence claim.

major comments (4)
  1. [§V.A, Table III; Abstract] The abstract claims reconstruction accuracy 'comparable' to COLMAP/GLOMAP. Table III contradicts this on Tanks and Temples: InstantSfM is below COLMAP and GLOMAP on every scene and every threshold (e.g., average AUC@5 = 75.63 vs. 78.48 for COLMAP and 78.61 for GLOMAP; Playground AUC@3 = 14.00 vs. 26.76). The authors attribute the deficit to using PyCeres for ViewGraphCalibration and OpenCV in Python for relative pose estimation. This attribution must be validated: please run the same pipeline with reference C++ Ceres/PoseLib front-ends (or with identical initializations) to show that the proposed GP/BA optimization core is not the cause, or revise the accuracy claim to describe the actual end-to-end system. As presented, the system does not deliver the claimed accuracy equivalence.
  2. [§IV.C, Eq. (4)] The depth-prior formulation is under-specified. Eq. (4) writes u_ij = v_ij - (X_j - t_i)/depth_ij, but the manuscript does not define depth_ij (Euclidean ray distance vs. z-depth from the depth map) nor the meaning of v_ij (unit ray direction, normalized pixel direction, or direction including intrinsics). If depth_ij is z-depth, the relationship (X_j - t_i)/depth_ij is not the same as the ray direction v_ij, and the residual in metric units is not well posed. The claim that the depth prior 'resolve[s] scale ambiguity' depends on this definition. Please give precise definitions and the depth-constrained Jacobian blocks for both the GP and BA stages.
  3. [§V.A, Tables I–II] On MipNeRF360 and DTU the accuracy evaluation is indirect: the paper reports only 3DGS rendering metrics (PSNR/SSIM/LPIPS), not camera pose or 3D point accuracy. Rendering quality can mask systematic SfM errors, and the unusual averages in Table I (e.g., Ours average PSNR 28.43 vs. COLMAP 26.92, driven by large drops in several COLMAP scenes) suggest evaluation protocol differences that are not explained. Please report direct SfM metrics (pose AUC or rotation/translation errors, triangulation reprojection errors) for these datasets, or at minimum explain why COLMAP/GLOMAP scores differ from their published behavior on the same data.
  4. [§V.A, Tables IV–V] The ScanNet and ScanNet++ depth-prior experiments use ground-truth depth from RGB-D sensors. This is an oracle external prior, so comparing 'Ours (w/ depth)' against COLMAP/GLOMAP (which receive no such prior) is not a fair comparison of SfM accuracy. The 'Ours (w/o depth)' ablation is useful, but the paper's statement that depth prior enables metric-scale reconstruction 'significantly outperform[ing] compared methods' should be clearly framed as a controlled study with an external oracle, not as a property of the SfM system alone. Please also report the metric-scale error with and without the depth prior explicitly, rather than only Chamfer distance.
minor comments (5)
  1. [§III.A, Eq. (1)] The notation v_ij is described as 'pixel ray direction vectors' but the equation uses 2D feature points inconsistently. Please clarify the exact definition and the dimension of v_ij (unit vector vs. pixel coordinate).
  2. [§IV, Algorithm 1] The sentence 'as shown in .' has an incomplete cross-reference; the algorithm should be cited as 'in Algorithm 1'.
  3. [§V.B, Fig. 3 and Fig. 4] The runtime plots are in log space, which obscures the actual speedup factors. Please include numeric labels or a table with exact timings for the key comparisons.
  4. [§V.B, Table VI] The acronym 'PBA' is not defined. Please expand it on first use, and clarify the hardware/software versions for Ceres-GPU and PBA.
  5. [References] Reference [26] (GLOMAP) is a key baseline but appears as a 2024 ECCV paper; the text uses GLOMAP in multiple places without a companion citation to the original arXiv version. Also, the related-work discussion of [15] (the sparse BA framework the method builds on) could more clearly state which components are inherited from [15] and which are new to this paper.

Circularity Check

0 steps flagged

No circular derivation: InstantSfM's efficiency and accuracy claims are empirically validated against external baselines; the depth prior is an external input, not a fitted prediction renamed as a result.

full rationale

The paper's central claims do not reduce to their own inputs. The efficiency claim is supported by direct runtime comparisons against COLMAP, GLOMAP, and GPU-BA baselines on standard benchmarks (Fig. 4, Tabs. VI–VII), not by a fitted parameter or by the cited prior work alone. The sparse-aware operations from [15] are an implementation dependency shared with the authors' prior work, but the present paper independently validates the resulting system with external timings and reconstruction metrics; this is not a load-bearing self-citation in the circular sense. The depth-prior mechanism (Sec. IV-C, Eq. 4) injects ground-truth or estimated depth maps as an external input; metric-scale outputs are a consequence of that external constraint, not a quantity fitted from the output and then called a prediction. All evaluation metrics (AUC, Chamfer, PSNR/SSIM/LPIPS) are computed externally relative to independent ground truth or baselines. The Tanks-and-Temples accuracy deficit (Tab. III) is an empirical inconsistency with the abstract's 'comparable accuracy' wording, but it is a correctness/fairness concern, not a circular derivation. No equation or contribution is defined in terms of the target result it is supposed to explain.

Axiom & Free-Parameter Ledger

2 free parameters · 4 axioms · 0 invented entities

InstantSfM introduces no new entities and no fitted global constants. Its load-bearing inputs are: (1) the external sparse-LM kernels of [15], (2) the GLOMAP GP objective, (3) external metric depth priors, and (4) an unspecified under-constrained-variable filter. The only optimizable per-observation quantity (d_ij) is precisely what the depth prior removes. The counts here reflect that the paper is an engineering integration rather than a derivation from first principles.

free parameters (2)
  • per-observation scale d_ij (Eq. 1, no-depth variant) = optimized per observation; values not reported
    In global positioning without depth prior, each 2D observation carries an optimizable scalar scale that absorbs ray depth; this is exactly the quantity the depth prior replaces (Sec. IV.C). Its presence without a prior leaves scale ambiguity, which is the paper's stated motivation.
  • under-constrained variable filtering rule (Sec. IV.D) = criterion not specified
    The numerical-stability mechanism drops Jacobian parameters that 'do not introduce meaningful gradient' in the forward pass. The threshold/criterion is never quantified, making the stability behavior effectively hand-chosen and irreproducible.
axioms (4)
  • domain assumption Sparse-aware LM primitives of [15] are correct, stable, and fast as claimed
    The whole pipeline (Fig. 2, Sec. IV.D) relies on [15]'s JTJ, JT r, and diag(A) sparse kernels; no derivation, convergence analysis, or independent verification is provided in this paper, and [15] shares authorship with this paper.
  • domain assumption GP objective (Eq. 1) from GLOMAP [26] with fixed rotations is a sound initialization for subsequent BA
    Adopted wholesale from [26]; the paper's Tanks-and-Temples degradation (intrinsics-translation ambiguity, up to ~25% intrinsic deviation) shows that this stage, combined with the PyCeres/OpenCV implementation, limits final accuracy.
  • domain assumption Provided depth maps are metric and view-aligned (depth_ij in Eq. 4)
    The metric-scale claim (contribution 3) holds only if the injected depth is true metric depth. ScanNet experiments use ground-truth RGB-D depth; if a user supplies scale-ambiguous monocular depth, 'resolving scale ambiguity' does not follow.
  • standard math Standard Levenberg-Marquardt convergence theory (Eq. 3 with diagonal damping)
    Uses the textbook LM update with diagonal damping; unremarkable but not proved or numerically analyzed here.

pith-pipeline@v1.3.0-alltime-deepseek · 12472 in / 19723 out tokens · 162674 ms · 2026-08-04T09:46:31.419156+00:00 · methodology

0 comments
read the original abstract

Structure-from-Motion (SfM) is a fundamental technique for recovering camera poses and scene structure from multi-view imagery, serving as a critical upstream component for applications ranging from 3D reconstruction to modern neural scene representations such as 3D Gaussian Splatting. However, most mature SfM systems remain CPU-centric and built upon traditional optimization toolchains, creating a growing mismatch with modern GPU-based, learning-driven pipelines and limiting scalability in large-scale scenes. While recent advances in GPU-accelerated bundle adjustment (BA) have demonstrated the potential of parallel sparse optimization, extending these techniques to build a complete global SfM system remains challenging due to unresolved issues in metric scale recovery and numerical robustness. In this paper, we implement a fully GPU-based and PyTorch-compatible global SfM system, named InstantSfM, to integrate seamlessly with modern learning pipelines. InstantSfM embeds metric depth priors directly into both global positioning and BA through a depth-constrained Jacobian structure, thereby resolving scale ambiguity within the optimization framework. To ensure numerical stability, we employ explicit filtering of under-constrained variables for the Jacobian matrix in an optimized GPU-friendly manner. Extensive experiments on diverse datasets demonstrate that InstantSfM achieves state-of-the-art efficiency while maintaining reconstruction accuracy comparable to both established classical pipelines and recent learning-based methods, showing up to ${\sim40\times}$ speedup over COLMAP on large-scale scenes.

Figures

Figures reproduced from arXiv: 2510.13310 by Chen Wang, Haozhe Lou, Jiageng Mao, Jiankun Zhong, Quankai Gao, Ulrich Neumann, Yue Wang, Zitong Zhan, Ziyu Chen.

Figure 1
Figure 1. Figure 1: Qualitative results on various datasets. Camera frustums as red are visualized for illustrating the estimated camera poses. Input images for each [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: We illustrate the sparse representation of Jacobian [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Comparisons of total SfM running time among COLMAP, GLOMAP [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Comparisons of BA and GP running time between GLOMAP and [PITH_FULL_IMAGE:figures/full_fig_p007_4.png] view at source ↗

discussion (0)

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

Forward citations

Cited by 2 Pith papers

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

  1. EPO: Boosting 3D Foundation Models with Edge-based Pose Optimization

    cs.CV 2026-07 unverdicted novelty 7.0

    EPO is a trackless, edge-map-alignment framework that refines pose estimates from 3D foundation models and matches or exceeds bundle-adjustment performance with substantially lower runtime and memory use.

  2. CSS-BA: Gate-Guided Column Space Search for Bundle Adjustment

    cs.CV 2026-07 conditional novelty 6.0

    A gate-guided subspace restriction of the Schur camera update improves pose accuracy in weak-geometry bundle adjustment without changing the objective.

Reference graph

Works this paper leans on

52 extracted references · 7 linked inside Pith · cited by 2 Pith papers

  1. [1]

    Building rome in a day,

    S. Agarwal, Y . Furukawa, N. Snavely, I. Simon, B. Curless, S. M. Seitz, and R. Szeliski, “Building rome in a day,”Communications of the ACM, vol. 54, no. 10, pp. 105–112, 2011

    2011

  2. [2]

    Building rome on a cloudless day,

    J.-M. Frahm, P. Fite-Georgel, D. Gallup, T. Johnson, R. Raguram, C. Wu, Y .-H. Jen, E. Dunn, B. Clipp, S. Lazebniket al., “Building rome on a cloudless day,” inComputer Vision–ECCV 2010: 11th European Conference on Computer Vision, Heraklion, Crete, Greece, September 5-11, 2010, Proceedings, Part IV 11. Springer, 2010, pp. 368–381

    2010

  3. [3]

    Stereo from uncalibrated cameras

    R. I. Hartley, R. Gupta, and T. Chang, “Stereo from uncalibrated cameras.” inCVPR, vol. 92, 1992, pp. 761–764

    1992

  4. [4]

    MacDonald,Digital heritage

    L. MacDonald,Digital heritage. Routledge, 2006

    2006

  5. [5]

    Om- nire: Omni urban scene reconstruction,

    Z. Chen, J. Yang, J. Huang, R. de Lutio, J. M. Esturo, B. Ivanovic, O. Litany, Z. Gojcic, S. Fidler, M. Pavone, L. Song, and Y . Wang, “Om- nire: Omni urban scene reconstruction,” inThe Thirteenth International Conference on Learning Representations, 2025

    2025

  6. [6]

    Megasam: Accurate, fast, and robust structure and motion from casual dynamic videos,

    Z. Li, R. Tucker, F. Cole, Q. Wang, L. Jin, V . Ye, A. Kanazawa, A. Holynski, and N. Snavely, “Megasam: Accurate, fast, and robust structure and motion from casual dynamic videos,”arXiv preprint arXiv:2412.04463, 2024

    Pith/arXiv arXiv 2024

  7. [7]

    Robo-GS: A physics consistent spatial-temporal model for robotic arm with hybrid representation,

    H. Lou, Y . Liu, Y . Pan, Y . Geng, J. Chen, W. Ma, C. Li, L. Wang, H. Feng, L. Shi, L. Luo, and Y . Shi, “Robo-GS: A physics consistent spatial-temporal model for robotic arm with hybrid representation,” in IEEE International Conference on Robotics and Automation (ICRA),

  8. [8]

    Nerf: Representing scenes as neural radiance fields for view synthesis,

    B. Mildenhall, P. P. Srinivasan, M. Tancik, J. T. Barron, R. Ramamoor- thi, and R. Ng, “Nerf: Representing scenes as neural radiance fields for view synthesis,”Communications of the ACM, vol. 65, no. 1, pp. 99–106, 2021

    2021

  9. [9]

    3d gaussian splatting for real-time radiance field rendering

    B. Kerbl, G. Kopanas, T. Leimkühler, and G. Drettakis, “3d gaussian splatting for real-time radiance field rendering.”ACM Trans. Graph., vol. 42, no. 4, pp. 139–1, 2023

    2023

  10. [10]

    Kinectfusion: Real-time dense surface mapping and tracking,

    R. A. Newcombe, S. Izadi, O. Hilliges, D. Molyneaux, D. Kim, A. J. Davison, P. Kohi, J. Shotton, S. Hodges, and A. Fitzgibbon, “Kinectfusion: Real-time dense surface mapping and tracking,” in2011 10th IEEE international symposium on mixed and augmented reality. Ieee, 2011, pp. 127–136

    2011

  11. [11]

    iMatching: Imperative correspondence learning,

    Z. Zhan, D. Gao, Y .-J. Lin, Y . Xia, and C. Wang, “iMatching: Imperative correspondence learning,” inEuropean Conference on Computer Vision (ECCV), 2024. [Online]. Available: https://arxiv.org/pdf/2312.02141.pdf

    Pith/arXiv arXiv 2024

  12. [12]

    Bundle adjustment in the large,

    S. Agarwal, N. Snavely, S. M. Seitz, and R. Szeliski, “Bundle adjustment in the large,” inComputer Vision–ECCV 2010: 11th European Conference on Computer Vision, Heraklion, Crete, Greece, September 5-11, 2010, Proceedings, Part II 11. Springer, 2010, pp. 29–42

    2010

  13. [13]

    Robust global translations with 1dsfm,

    K. Wilson and N. Snavely, “Robust global translations with 1dsfm,” in European conference on computer vision. Springer, 2014, pp. 61–75

    2014

  14. [14]

    Robust bundle adjustment revisited,

    C. Zach, “Robust bundle adjustment revisited,” inComputer Vision – ECCV 2014, D. Fleet, T. Pajdla, B. Schiele, and T. Tuytelaars, Eds. Cham: Springer International Publishing, 2014, pp. 772–787

    2014

  15. [15]

    Bundle adjustment in the eager mode,

    Z. Zhan, H. Xu, Z. Fang, X. Wei, Y . Hu, and C. Wang, “Bundle adjustment in the eager mode,”arXiv preprint arXiv:2409.12190, 2024

    Pith/arXiv arXiv 2024

  16. [16]

    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

  17. [17]

    Superpoint: Self- supervised interest point detection and description,

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

    2018

  18. [18]

    Loftr: Detector- free local feature matching with transformers,

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

    2021

  19. [19]

    Efficient loftr: Semi- dense local feature matching with sparse-like speed,

    Y . Wang, X. He, S. Peng, D. Tan, and X. Zhou, “Efficient loftr: Semi- dense local feature matching with sparse-like speed,” inProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2024, pp. 21 666–21 675

    2024

  20. [20]

    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,” inProceed- ings of the IEEE/CVF conference on computer vision and pattern recognition, 2020, pp. 4938–4947

    2020

  21. [21]

    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

    1981

  22. [22]

    Triangulation,

    R. I. Hartley and P. Sturm, “Triangulation,”Computer vision and image understanding, vol. 68, no. 2, pp. 146–157, 1997

    1997

  23. [23]

    Structure-from-motion revisited,

    J. L. Schönberger and J.-M. Frahm, “Structure-from-motion revisited,” inConference on Computer Vision and Pattern Recognition (CVPR), 2016

    2016

  24. [24]

    Bundler: Structure from motion (sfm) for unordered image collections,

    N. Snavely, “Bundler: Structure from motion (sfm) for unordered image collections,” Code available at http://phototour.cs.washington.edu/bundler/, 2010

    2010

  25. [25]

    LaMAR: Benchmarking Localization and Mapping for Augmented Reality,

    P.-E. Sarlin, M. Dusmanu, J. L. Schönberger, P. Speciale, L. Gruber, V . Larsson, O. Miksik, and M. Pollefeys, “LaMAR: Benchmarking Localization and Mapping for Augmented Reality,” inECCV, 2022

    2022

  26. [26]

    Global structure-from-motion revisited,

    L. Pan, D. Baráth, M. Pollefeys, and J. L. Schönberger, “Global structure-from-motion revisited,” inEuropean Conference on Computer Vision. Springer, 2024, pp. 58–77

    2024

  27. [27]

    A global linear method for camera pose registration,

    N. Jiang, Z. Cui, and P. Tan, “A global linear method for camera pose registration,” inProceedings of the IEEE international conference on computer vision, 2013, pp. 481–488

    2013

  28. [28]

    Global fusion of relative motions for robust, accurate and scalable structure from motion,

    P. Moulon, P. Monasse, and R. Marlet, “Global fusion of relative motions for robust, accurate and scalable structure from motion,” in Proceedings of the IEEE international conference on computer vision, 2013, pp. 3248–3255

    2013

  29. [29]

    Linear multiview reconstruction of points, lines, planes and cameras using a reference plane,

    Rother, “Linear multiview reconstruction of points, lines, planes and cameras using a reference plane,” inProceedings Ninth IEEE International Conference on Computer Vision. IEEE, 2003, pp. 1210– 1217

    2003

  30. [30]

    Vggsfm: Visual geometry grounded deep structure from motion,

    J. Wang, N. Karaev, C. Rupprecht, and D. Novotny, “Vggsfm: Visual geometry grounded deep structure from motion,” inProceedings of the IEEE/CVF conference on computer vision and pattern recognition, 2024, pp. 21 686–21 697

    2024

  31. [31]

    Continuous 3d perception model with persistent state,

    Q. Wang, Y . Zhang, A. Holynski, A. A. Efros, and A. Kanazawa, “Continuous 3d perception model with persistent state,”arXiv preprint arXiv:2501.12387, 2025

    Pith/arXiv arXiv 2025

  32. [32]

    Vggt: Visual geometry grounded transformer,

    J. Wang, M. Chen, N. Karaev, A. Vedaldi, C. Rupprecht, and D. Novotny, “Vggt: Visual geometry grounded transformer,” inPro- ceedings of the Computer Vision and Pattern Recognition Conference, 2025, pp. 5294–5306

    2025

  33. [33]

    Dust3r: Geometric 3d vision made easy,

    S. Wang, V . Leroy, Y . Cabon, B. Chidlovskii, and J. Revaud, “Dust3r: Geometric 3d vision made easy,” inProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2024, pp. 20 697–20 709

    2024

  34. [34]

    The levenberg-marquardt algorithm,

    A. Ranganathan, “The levenberg-marquardt algorithm,”Tutoral on LM algorithm, vol. 11, no. 1, pp. 101–110, 2004

    2004

  35. [35]

    Ceres solver: Tutorial & reference,

    S. Agarwal, K. Mierleet al., “Ceres solver: Tutorial & reference,” Google Inc, vol. 2, no. 72, p. 8, 2012

    2012

  36. [36]

    borglab/gtsam

    F. Dellaert and G. Contributors, “borglab/gtsam”,” 2022

    2022

  37. [37]

    Deeplm: Large-scale nonlinear least squares on deep learning frameworks using stochastic domain decomposition,

    J. Huang, S. Huang, and M. Sun, “Deeplm: Large-scale nonlinear least squares on deep learning frameworks using stochastic domain decomposition,” inProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2021, pp. 10 308–10 317

    2021

  38. [38]

    Ceres Solver,

    S. Agarwal, K. Mierle, and The Ceres Solver Team, “Ceres Solver,” Oct

  39. [39]

    Building rome with convex optimization,

    H. Han and H. Yang, “Building rome with convex optimization,”arXiv preprint arXiv:2502.04640, 2025

    Pith/arXiv arXiv 2025

  40. [40]

    Rotation averaging,

    R. Hartley, J. Trumpf, Y . Dai, and H. Li, “Rotation averaging,” International journal of computer vision, vol. 103, pp. 267–305, 2013

    2013

  41. [41]

    Combining two-view constraints for motion estima- tion,

    V . M. Govindu, “Combining two-view constraints for motion estima- tion,” inProceedings of the 2001 IEEE computer society conference on computer vision and pattern recognition. CVPR 2001, vol. 2. IEEE, 2001, pp. II–II

    2001

  42. [42]

    Robust estimation of a location parameter,

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

    1992

  43. [43]

    Mip-nerf 360: Unbounded anti-aliased neural radiance fields,

    J. T. Barron, B. Mildenhall, D. Verbin, P. P. Srinivasan, and P. Hedman, “Mip-nerf 360: Unbounded anti-aliased neural radiance fields,” in Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, 2022, pp. 5470–5479

    2022

  44. [44]

    gsplat: An open-source library for Gaussian splatting,

    V . Ye, R. Li, J. Kerr, M. Turkulainen, B. Yi, Z. Pan, O. Seiskari, J. Ye, J. Hu, M. Tancik, and A. Kanazawa, “gsplat: An open-source library for Gaussian splatting,”arXiv preprint arXiv:2409.06765, 2024. [Online]. Available: https://arxiv.org/abs/2409.06765

    Pith/arXiv arXiv 2024

  45. [45]

    Detector-free structure from motion,

    X. He, J. Sun, Y . Wang, S. Peng, Q. Huang, H. Bao, and X. Zhou, “Detector-free structure from motion,” inProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2024, pp. 21 594–21 603

    2024

  46. [46]

    Scannet++: A high- fidelity dataset of 3d indoor scenes,

    C. Yeshwanth, Y .-C. Liu, M. Nießner, and A. Dai, “Scannet++: A high- fidelity dataset of 3d indoor scenes,” inProceedings of the International Conference on Computer Vision (ICCV), 2023

    2023

  47. [47]

    Large scale multi-view stereopsis evaluation,

    R. Jensen, A. Dahl, G. V ogiatzis, E. Tola, and H. Aanaes, “Large scale multi-view stereopsis evaluation,” inProceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2014

    2014

  48. [48]

    Large-scale data for multiple-view stereopsis,

    H. Aanæs, R. R. Jensen, G. V ogiatzis, E. Tola, and A. B. Dahl, “Large-scale data for multiple-view stereopsis,”International Journal of Computer Vision, pp. 1–16, 2016

    2016

  49. [49]

    Tanks and temples: Benchmarking large-scale scene reconstruction,

    A. Knapitsch, J. Park, Q.-Y . Zhou, and V . Koltun, “Tanks and temples: Benchmarking large-scale scene reconstruction,”ACM Transactions on Graphics, vol. 36, no. 4, 2017

    2017

  50. [50]

    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,” inProc. Computer Vision and Pattern Recognition (CVPR), IEEE, 2017

    2017

  51. [2023]

    Available: https://github.com/ceres-solver/ceres-solver

    [Online]. Available: https://github.com/ceres-solver/ceres-solver

  52. [2025]

    Available: https://arxiv.org/abs/2408.14873

    [Online]. Available: https://arxiv.org/abs/2408.14873

    Pith/arXiv arXiv