Pith. sign in

REVIEW 5 major objections 5 minor 60 references

C-DOG: Multi-View Multi-instance Feature Association Using Connected {\delta}-Overlap Graphs

T0 review · 5 major / 5 minor · reviewed 2026-08-06 · deepseek-v4-flash

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

Pith's one-line read C-DOG claims that a connected delta-overlap graph built from epipolar distances alone can associate 2D detections of identical objects across multiple calibrated views, outperforming geometry-based baselines under noise, high object…

desk verdict C-DOG is a plausible featureless association pipeline with a real gap: its argmin initialization can permanently drop true correspondences, and the evaluation never measures that. read the letter →

arxiv 2507.14095 v2 pith:4SXP2E6R submitted 2025-07-18 cs.CV

Yung-Hong Sun , Ting-Hung Lin , Jiangang Chen , Hongrui Jiang , Yu Hen Hu This is my paper · ORCID
classification cs.CV
keywords multi-viewassociationepipolargeometryδ-overlapgraphmulti-instancematchingfeatureless3Dreconstructionclusteringoutlierrejection
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

Multi-view 3D reconstruction usually begins by matching visual features across images, but scenes full of identical objects offer appearance no way to tell one instance from another. The paper's claim is that geometry alone can do the association: given only 2D point coordinates and known camera poses, a graph whose edges are minimal epipolar distances, pruned by a $\delta$-neighborhood-overlap rule and then by IQR-filtered back-projection error, groups the views' detections into the correct 3D instances. On a synthetic benchmark with up to 130 randomly placed 3D points observed by ten calibrated cameras under Gaussian noise up to 5 pixels, C-DOG reports higher group-level, point-level, and perfect-group F1/IoU than geometry-based baselines, and degrades more slowly as object count and noise grow. A sympathetic reader should care because a working feature-free associator would remove the main bottleneck in texture-poor, privacy-sensitive, or repetitively structured scenes, and would need no training or learned descriptors.

What carries the argument

The connected $\delta$-overlap graph. Each 2D detection is a node; candidate edges are view-wise minimal epipolar-distance pairs below threshold $\tau$. The load-bearing mechanism is the $\delta$-overlap criterion: for an edge to survive, the closed neighborhoods of its two endpoints must share at least a fraction $\delta$ (chosen 0.5–0.55 by grid search) of the larger neighborhood, measured with the Szymkiewicz–Simpson coefficient. This turns pairwise geometry into a structural vote—true within-group edges are backed by many mutual neighbors, while spurious bridges between dense clusters score low. The second mechanism is IQR-based pruning on per-node average back-projection error, which removes outliers that satisfy epipolar and overlap constraints but triangulate inconsistently.

What would settle it

Take a two-view setup with two 3D points whose projections land almost on the same epipolar line, add Gaussian noise with $\sigma=3$ pixels, and run the initialization of Algorithm 1. Because the initialization keeps only the argmin point in each other view below $\tau$, whenever the wrong point is closer the true edge is absent from the graph; measuring the fraction of true edges present after initialization against the same fraction when all edges below $\tau$ are kept would decide whether the argmin premise holds. If the final group recall stays high only in the first case, the method depends on the wrong point rarely being closer.

Watch

Extended reading notes

Core claim

On the paper's own terms, the discovery is that structural consensus in a connected $\delta$-overlap graph recovers what raw epipolar scores cannot: under noise, correct and incorrect pairwise associations become nearly indistinguishable, but the correct associations are the ones embedded in a dense, mutually agreeing neighborhood, and $\delta$-overlap clustering isolates exactly that structure. The algorithm keeps, for each point and each other view, only the single candidate with minimum epipolar distance below $\tau = \alpha\sqrt{2}\sigma$, weights each retained edge by its epipolar distance, and then prunes any edge whose Szymkiewicz–Simpson neighborhood overlap falls below $\delta$, so that a group of nodes survives only if its members agree on their neighbors. Group-level IQR filtering on triangulated back-projection error removes residual outliers the graph structure cannot see, and a final stage drops whole groups whose error jumps abruptly. The paper reports the full model reaching, for example, G-F1 of 0.950, PG-F1 of 0.937, and mP-F1 of 0.904 at $\sigma=0$, and 0.881, 0.761, and 0.727 at $\sigma=3$ on the 10-view benchmark, outperforming all geometry-based baselines listed in its comparison tables.

Load-bearing premise

The load-bearing premise is that, for each 2D point and each other view, the true matching point has the smallest epipolar distance below the threshold $\tau$; if noise or a nearby object places a wrong point closer on the epipolar line, that true edge is never added and the later overlap and IQR stages cannot recover it.

Editorial extensions

If this is right

  • C-DOG can serve as a training-free correspondence module between multi-view object detection and 3D reconstruction, since it consumes only 2D coordinates and camera poses and never uses appearance descriptors.
  • In calibrated multi-camera rigs, such as the 10-camera laparoscopic training setup that motivated the paper, redundant views act as a voting mechanism: beyond roughly five views, additional cameras improve group association despite adding noisy pairwise scores.
  • The explicit noise model $\tau=\alpha\sqrt{2}\sigma$ gives practitioners a principled way to set the association threshold from an estimated pixel-noise level instead of a learned hyperparameter.
  • The IQR and back-projection stage is load-bearing for precision under noise: comparing the full model with the no-IQR variant in the paper's tables shows most of the point-level and perfect-group gains vanish without it.
  • Because the benchmark deliberately omits texture, the reported gains isolate geometric association quality, so a positive result implies the method suits texture-poor, privacy-sensitive, or repeated-object scenes.

Reading between the lines

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

  • Beyond the paper, the argmin initialization sets a ceiling: the method can only recover a true correspondence if that point is the closest epipolar candidate in each other view, so scenes with near-collinear 3D points or a biased camera calibration would break it before the pruning stages act.
  • Beyond the paper, the $\delta$-overlap criterion could be reused as a generic consistency filter for learned matchers, cleaning soft graph-neural-network correspondences by requiring mutual neighborhood agreement; this is a testable extension the paper does not run.
  • Beyond the paper, since the synthetic benchmark adds Gaussian noise to clean projections and does not simulate missing detections or false positives, the robustness claims cover localization noise; performance under occlusion and detector dropouts remains an open question.
  • Beyond the paper, the reported polynomial runtime growth with instance count suggests that spatial hashing or hierarchical view selection would be needed before C-DOG scales to real-time use with hundreds of objects.
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

5 major / 5 minor

Summary. The paper proposes C-DOG, a multi-view, multi-instance feature association method that uses only epipolar geometry and known camera poses, with no appearance descriptors. The pipeline builds an initial graph whose nodes are 2D detections and whose edges are candidate correspondences selected by per-view argmin epipolar distance below a threshold τ (Algorithm 1, Eq. 14); it then prunes weak edges using a Szymkiewicz–Simpson neighborhood-overlap score with threshold δ (Algorithm 2), removes intra-group outliers via IQR filtering of back-projection errors (Algorithm 3), and finally removes entire groups whose BPE rises abruptly (Section IV-F). The method is evaluated on a synthetic benchmark with up to 130 points, 10 views, and Gaussian noise up to σ = 5, where it reports higher group-level, point-level, and perfect-group F1 scores than several geometry-based baselines.

Significance. If the reported results hold, C-DOG is a useful training-free and feature-free association module for multi-view 3D reconstruction in texture-poor or repeated-instance scenes. The paper's strengths include a clearly described graph pipeline, extensive synthetic comparisons with multiple metrics, an ablation of view count, and a public code release. However, the current evidence is weakened by three load-bearing gaps: the benchmark omits missing detections and false positives; the key thresholds (δ, τ) are tuned on the evaluation benchmark or use the known noise level of the synthetic data; and the initialization's per-view argmin can permanently discard true correspondences, a failure mode that is never measured. The false statement that the fundamental matrix is skew-symmetric is also a serious mathematical error in the preliminaries, though it appears fixable. These issues make the paper's central claims plausible but not yet fully supported.

major comments (5)
  1. [Section III-A] The statement following Eq. (1) that the fundamental matrix is skew-symmetric, F_{m,m'} = -F_{m,m'}^T, is incorrect in general. A nonzero 3×3 skew-symmetric matrix has rank 2 and zero diagonal, but the fundamental matrix is not generally skew-symmetric; for example, the essential matrix E = [t]× R is not skew-symmetric unless R is identity (in a special coordinate frame). Moreover, the placement of intrinsic and extrinsic terms in Eq. (1) is not the standard form F = K'^{-T} [t]× R K^{-1}. Since the epipolar-line computations in Eqs. (2)–(6) all depend on F, the authors must correct this derivation or clearly state any special assumptions under which their formula and symmetry claim hold.
  2. [Algorithm 1 / Eq. (14)] The initialization retains, for each ordered pair of views and each point, only the single point with minimum epipolar distance below τ. If the true correspondent is not that argmin—which becomes increasingly likely as object density and noise grow—the true edge is never added, and no later stage (weak-edge pruning, IQR outlier removal, or group removal) ever inserts an edge. The paper does not report the fraction of ground-truth associations that survive this initialization, so the impact of this bottleneck on the reported F1 values is unknown. I request a direct measurement of the survival rate of true correspondences after Algorithm 1, as a function of σ and instance count, and a discussion of whether retaining all candidates below τ (rather than only the argmin) would improve recall without reducing precision.
  3. [Section IV-G] The benchmark description explicitly states that the dataset 'does not simulate missing detections or false positives and each 3D point has a corresponding 2D observation in all views.' Real detector outputs contain both missing detections and false positives, and the abstract claims robustness under 'restricted camera overlap,' yet the experiments only drop whole views rather than individual detections. These two forms of missing data are not equivalent, and the absence of false positives likely inflates precision and perfect-group scores. Please add experiments that randomly remove a fraction of 2D points and add spurious detections, or at least quantify how the reported F1 degrades under these conditions.
  4. [Sections IV-C and IV-D] The threshold τ is set using the true noise standard deviation σ of each synthetic test case (Eq. 16), and the overlap threshold δ is selected by grid search on the same benchmark used for evaluation ('Through grid-search on our benchmark, we found that setting δ ∈ [0.5, 0.55] yields the best overall performance'). This constitutes tuning on the test set, so the reported numbers are optimistic relative to a deployment scenario where σ is unknown and δ must be chosen without access to ground-truth labels. I suggest reserving a portion of the benchmark for tuning and reporting performance on a held-out set, or providing sensitivity curves over δ and τ with σ estimated from the observed 2D points rather than taken from the data-generation process.
  5. [Section IV-F] The 'Error Association Group Removal' step is described only in prose: groups are sorted by BPE and remaining groups are discarded when 'the difference between two successive BPE values exhibits a sudden increase.' There is no formal criterion, no pseudocode, no threshold, and no ablation isolating this step's contribution. Since this stage removes entire groups and directly affects every reported metric, it must be specified precisely (e.g., a gap statistic or percentile rule) and its effect on the final F1 should be measured, for example by comparing the full pipeline against a version without this stage.
minor comments (5)
  1. [Abstract] The abstract states that edges are retained 'where the epipolar distance is less than a threshold delta,' but in the body δ is the overlap threshold for weak-edge pruning while the epipolar threshold is τ (Eq. 14 and Eq. 16). This conflation of δ and τ should be corrected for clarity.
  2. [Section IV-C] The text says the initialization produces a 'bidirectional connection graph,' but Algorithm 1 adds directed edges (line 10) and the graph is then treated as undirected in the overlap computation. Please clarify whether edges are symmetrized before connected-component extraction and overlap scoring.
  3. [Section IV-G] The sentence 'A feature detection (and localization) algorithm will is used to detect...' contains a grammatical error and should be rewritten.
  4. [Section IV-D] The statement that δ ∈ [0.5, 0.55] 'is intuitive, as it requires that at least half of each node's neighborhood overlaps' is not fully justified because the denominator in Eq. (17) uses the maximum neighborhood size rather than the size of the smaller neighborhood; please clarify the interpretation of the threshold for asymmetric neighborhoods.
  5. [Section V-A] The definition of mP-R is ambiguous: point-level recall is computed per predicted group and averaged across groups, which may not equal the usual recall over all ground-truth points. Please state explicitly how false negatives are counted across groups.

Circularity Check

2 steps flagged · score 4.0 of 10

C-DOG's core derivation is independent, but its headline evaluation is partly self-referential: the δ pruning threshold is grid-searched on the same benchmark used for reporting, and the epipolar threshold is set from the benchmark's known noise σ.

  1. fitted input called prediction [Section IV-D (Algorithm 2, Weak Edge Pruning) and Section V-A/Tables II-V]
    "Through grid-search on our benchmark, we found that setting δ ∈ [0.5, 0.55] yields the best overall performance."

    This sentence reports that the weak-edge pruning threshold δ was selected by grid search on 'our benchmark' — the same synthetic benchmark whose results are later reported as the method's headline performance (Section V-A, Tables II-V). The claim that C-DOG 'outperforms geometry-based baseline algorithms' is therefore evaluated with a parameter chosen to maximize those exact performance scores on the test set. No held-out validation split or cross-validation is described, so the reported F1/IoU numbers are not an independent estimate of method quality; the evaluation is partly a restatement of the threshold selection.

  2. other [Section IV-C (Eq. 15-16) and Section IV-G (Synthetic Benchmark Generation)]
    "τ = α√2σ, (16) where α controls the tolerance to noise."

    The epipolar threshold τ is computed directly from σ, the standard deviation of the zero-mean Gaussian noise added to the 2D projections. In the synthetic benchmark, σ is the ground-truth generative parameter of the test data ('The noise standard deviation varies from 0 to 5 pixels'). Thus the method's operating point is set using information from the evaluation data's own corruption process. The paper states σ 'can be reliably estimated from 2D observations' but provides no experiment with an estimated σ, so the reported robustness holds only under the oracle condition that the exact test noise level is known. This is an evaluation-condition self-reference rather than a derivation-level circularity.

full rationale

The algorithmic derivation—epipolar distance thresholding, δ-overlap pruning, IQR outlier removal, and error-group removal—is not circular: each stage transforms the graph using stated geometric/statistical criteria, and the output association groups are not equal by construction to any fitted parameter. The method is self-contained against external geometry-based baselines in the sense that it runs without training. However, the experimental validation is partially self-referential because (1) the key pruning threshold δ is tuned on the same benchmark used to report results, and (2) the epipolar threshold uses the true noise σ of the test data. These two choices make the headline performance numbers partly a function of the evaluation set itself. The final error-association-group removal criterion is also under-specified ('sudden increase' is never quantified), which is a reproducibility concern, not a circularity. No load-bearing self-citation chain or imported uniqueness theorem is present. Score 4 reflects a central algorithm with independent content but a benchmark-tuned evaluation.

Assumptions & free parameters 4 free parameters · 6 assumptions · 0 invented entities

The central claim rests on calibrations being accurate, a Gaussian noise model with known sigma, dense-clique structure of true associations, and a synthetic benchmark that excludes missing detections and false positives. The free parameters (delta, alpha values, and the unspecified BPE gap threshold) are tuned or chosen without a separate validation split, which weakens the claim that the method generalizes beyond the benchmark.

free parameters (4)
  • delta (neighborhood overlap threshold) = 0.5 to 0.55 (grid-searched)
    Section IV-D: 'Through grid-search on our benchmark, we found that setting delta in [0.5, 0.55] yields the best overall performance.' The same benchmark is used for final evaluation, so the threshold is tuned to the test distribution.
  • alpha for epipolar threshold tau = 2
    Section IV-C Eq (16): tau = alpha * sqrt(2) * sigma with alpha = 2. In experiments sigma is the known noise level added to the synthetic data; in practice sigma must be estimated.
  • IQR multiplier alpha = 2
    Section IV-E Eq (19): alpha = 2 is chosen for outlier rejection strictness without ablation.
  • BPE gap threshold for error group removal = unspecified
    Section IV-F: groups are discarded after a 'sudden increase' between successive sorted average BPE values, but no quantitative criterion is given.
assumptions (6)
  • domain assumption Camera intrinsics and extrinsics are known and calibrated
    Section IV-A problem formulation assumes M calibrated cameras.
  • standard math Pinhole camera model and epipolar geometry hold
    Section III-A uses the fundamental matrix and epipolar lines from Szeliski.
  • domain assumption 2D observation noise is zero-mean Gaussian with known standard deviation sigma
    Section IV-C Eq (15) models noise as N(0, sigma^2); experiments inject this noise and use sigma in tau.
  • domain assumption True correspondences form dense cliques in the association graph
    Section IV-D justifies neighborhood-overlap pruning by assuming correct groups are densely connected.
  • domain assumption Per-node average BPE distributions are informative for outlier detection via IQR
    Section IV-E Algorithm 3 rejects nodes whose mean BPE lies outside the IQR bounds.
  • domain assumption Synthetic scenes contain no missing detections or false positives; every 3D point is visible in all retained views
    Section IV-G states the benchmark does not simulate missing detections or false positives.

how reviews work

0 comments
Cite this review

Pith. "Pith review of C-DOG: Multi-View Multi-instance Feature Association Using Connected {\delta}-Overlap Graphs." pith.science (2026). https://pith.science/paper/4SXP2E6R

@misc{pith2026250714095,
  author       = {Pith},
  title        = {Pith review of: C-DOG: Multi-View Multi-instance Feature Association Using Connected \delta-Overlap Graphs},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/4SXP2E6R}},
  note         = {Machine review of arXiv:2507.14095}
}
read the original abstract

Multi-view multi-instance feature association constitutes a crucial step in 3D reconstruction, facilitating the consistent grouping of object instances across various camera perspectives. The presence of multiple identical objects within a scene often leads to ambiguities for appearance-based feature matching algorithms. Our work circumvents this challenge by exclusively employing geometrical constraints, specifically epipolar geometry, for feature association. We introduce C-DOG (Connected delta-Overlap Graph), an algorithm designed for robust geometrical feature association, even in the presence of noisy feature detections. In a C-DOG graph, two nodes representing 2D feature points from distinct views are connected by an edge if they correspond to the same 3D point. Each edge is weighted by its epipolar distance. Ideally, true associations yield a zero distance; however, noisy feature detections can result in non-zero values. To robustly retain edges where the epipolar distance is less than a threshold delta, we employ a Szymkiewicz--Simpson coefficient. This process leads to a delta-neighbor-overlap clustering of 2D nodes. Furthermore, unreliable nodes are pruned from these clusters using an Inter-quartile Range (IQR)-based criterion. Our extensive experiments on synthetic benchmarks demonstrate that C-DOG not only outperforms geometry-based baseline algorithms but also remains remarkably robust under demanding conditions. This includes scenes with high object density, no visual features, and restricted camera overlap, positioning C-DOG as an excellent solution for scalable 3D reconstruction in practical applications.

Figures

Figures reproduced from arXiv: 2507.14095 by the authors.

Figure 1
Figure 1. Demonstration of C-DOG applied to 130 3D points within a multi-camera system from our benchmark. (a) Multiple visually indistinguishable 3D points are observed within a multi-camera setup. (b) Each camera captures 2D projections of these 3D points, which can be obtained using object detectors or keypoint estimators. (c) C-DOG is employed to establish correspondences across views by associating multi-view multi-objec… view at source ↗
Figure 2
Figure 2. Example of multi-view multi-object association with two views and [PITH_FULL_IMAGE:figures/full_fig_p002_2.png] view at source ↗
Figure 3
Figure 3. Illustration of the epipolar line concept. Views [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗
Figures from the paper (6 more)
Figure 4
Figure 4. Figure 4: Overview of C-DOG with an illustrative example with three 3D points and 6 cameras: (a) A multi-camera system observes object points from multiple [PITH_FULL_IMAGE:figures/full_fig_p005_4.png]
Figure 5
Figure 5. Figure 5: An example of association graph initialization using object instances [PITH_FULL_IMAGE:figures/full_fig_p007_5.png]
Figure 6
Figure 6. Figure 6: Our method consistently outperformed the baselines in [PITH_FULL_IMAGE:figures/full_fig_p010_6.png]
Figure 6
Figure 6. Figure 6: Accuracy comparison of various baselines across varying noise levels and object instance counts. [PITH_FULL_IMAGE:figures/full_fig_p012_6.png]
Figure 7
Figure 7. Figure 7: Comparison of 3D reconstruction results for 130 instances across various baselines under different noise levels. (a) 10 observision views with visualized [PITH_FULL_IMAGE:figures/full_fig_p013_7.png]
Figure 8
Figure 8. Figure 8: Comparison of time cost across different methods under varying object [PITH_FULL_IMAGE:figures/full_fig_p014_8.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

60 extracted references · 54 canonical work pages

  1. [1]

    Multi-view stereo: A tutorial,

    Y . Furukawa, C. Hern ´andez, et al. , “Multi-view stereo: A tutorial,” Foundations and trends® in Computer Graphics and Vision , vol. 9, no. 1-2, pp. 1–148, 2015

    work page 2015

  2. [2]

    Multiview imaging and 3dtv,

    A. Kubota, A. Smolic, M. Magnor, M. Tanimoto, T. Chen, and C. Zhang, “Multiview imaging and 3dtv,” IEEE signal processing magazine , vol. 24, no. 6, pp. 10–21, 2007

    work page 2007

  3. [3]

    Lifting multi- view detection and tracking to the bird’s eye view,

    T. Teepe, P. Wolters, J. Gilg, F. Herzog, and G. Rigoll, “Lifting multi- view detection and tracking to the bird’s eye view,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp. 667–676, 2024

    work page 2024

  4. [4]

    For 3d laparoscopy: a step toward advanced surgical navigation: how to get maximum benefit from 3d vision,

    W. Kunert, P. Storz, and A. Kirschniak, “For 3d laparoscopy: a step toward advanced surgical navigation: how to get maximum benefit from 3d vision,” Surgical endoscopy, vol. 27, no. 2, pp. 696–699, 2013

    work page 2013

  5. [5]

    The slam problem: a survey,

    J. Aulinas, Y . Petillot, J. Salvi, and X. Llad ´o, “The slam problem: a survey,” Artificial Intelligence Research and Development, pp. 363–371, 2008

    work page 2008

  6. [6]

    A survey of feature matching methods,

    Q. Huang, X. Guo, Y . Wang, H. Sun, and L. Yang, “A survey of feature matching methods,” IET Image Processing , vol. 18, no. 6, pp. 1385– 1410, 2024

    work page 2024

  7. [7]

    Szeliski, Computer vision: algorithms and applications

    R. Szeliski, Computer vision: algorithms and applications . Springer Nature, 2022

    2022

  8. [8]

    View-centric multi-object tracking with homographic matching in moving uav,

    D. Ji, S. Gao, L. Zhu, Q. Zhu, Y . Zhao, P. Xu, H. Lu, F. Zhao, and J. Ye, “View-centric multi-object tracking with homographic matching in moving uav,” arXiv preprint arXiv:2403.10830 , 2024

    work page arXiv 2024

Show all 60 references
  1. [9]

    Collaborative association network for multi-view multi-human association and tracking using constraint optimization and object search,

    Z. Zhang, F. Shi, C. Jia, M. Zhao, and X. Cheng, “Collaborative association network for multi-view multi-human association and tracking using constraint optimization and object search,” in ICASSP 2025- 2025 IEEE International Conference on Acoustics, Speech and Signal Processi...

    2025

  2. [10]

    Adaptive multi-view and temporal fusing transformer for 3d human pose estimation,

    H. Shuai, L. Wu, and Q. Liu, “Adaptive multi-view and temporal fusing transformer for 3d human pose estimation,” IEEE Transactions on Pattern Analysis and Machine Intelligence , vol. 45, no. 4, pp. 4122– 4135, 2022

    2022

  3. [11]

    Direct multi-view multi- person 3d pose estimation,

    J. Zhang, Y . Cai, S. Yan, J. Feng, et al. , “Direct multi-view multi- person 3d pose estimation,” Advances in Neural Information Processing Systems, vol. 34, pp. 13153–13164, 2021

    2021

  4. [12]

    Efficient hierar- chical multi-view fusion transformer for 3d human pose estimation,

    K. Zhou, L. Zhang, F. Lu, X.-D. Zhou, and Y . Shi, “Efficient hierar- chical multi-view fusion transformer for 3d human pose estimation,” in Proceedings of the 31st ACM International Conference on Multimedia , pp. 7512–7520, 2023

    2023

  5. [13]

    Ppt: token-pruned pose transformer for monocular and multi-view human pose estimation,

    H. Ma, Z. Wang, Y . Chen, D. Kong, L. Chen, X. Liu, X. Yan, H. Tang, and X. Xie, “Ppt: token-pruned pose transformer for monocular and multi-view human pose estimation,” in European Conference on Computer Vision, pp. 424–442, Springer, 2022

    2022

  6. [14]

    Adafuse: Adaptive multiview fusion for accurate human pose estimation in the wild,

    Z. Zhang, C. Wang, W. Qiu, W. Qin, and W. Zeng, “Adafuse: Adaptive multiview fusion for accurate human pose estimation in the wild,” International Journal of Computer Vision , vol. 129, pp. 703–718, 2021

    2021

  7. [15]

    Towards discriminative representation: Multi- view trajectory contrastive learning for online multi-object tracking,

    E. Yu, Z. Li, and S. Han, “Towards discriminative representation: Multi- view trajectory contrastive learning for online multi-object tracking,” in Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, pp. 8834–8843, 2022

    2022

  8. [16]

    Shape-aware multi- person pose estimation from multi-view images,

    Z. Dong, J. Song, X. Chen, C. Guo, and O. Hilliges, “Shape-aware multi- person pose estimation from multi-view images,” in Proceedings of the IEEE/CVF International Conference on Computer Vision , pp. 11158– 11168, 2021

    2021

  9. [17]

    Messytable: Instance association in multiple camera views,

    Z. Cai, J. Zhang, D. Ren, C. Yu, H. Zhao, S. Yi, C. K. Yeo, and C. Change Loy, “Messytable: Instance association in multiple camera views,” in Computer Vision–ECCV 2020: 16th European Conference, Glasgow, UK, August 23–28, 2020, Proceedings, Part XI 16 , pp. 1–16, Springer, 2020

    2020

  10. [18]

    Fast and robust multi-person 3d pose estimation from multiple views,

    J. Dong, W. Jiang, Q. Huang, H. Bao, and X. Zhou, “Fast and robust multi-person 3d pose estimation from multiple views,” in Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , pp. 7792–7801, 2019

    2019

  11. [19]

    Near-optimal joint object matching via convex relaxation,

    Y . Chen, L. J. Guibas, and Q.-X. Huang, “Near-optimal joint object matching via convex relaxation,” arXiv preprint arXiv:1402.1473, 2014

    2014 arXiv

  12. [20]

    Clear: A consistent lifting, embedding, and alignment rectification algorithm for multiview data association,

    K. Fathian, K. Khosoussi, Y . Tian, P. Lusk, and J. P. How, “Clear: A consistent lifting, embedding, and alignment rectification algorithm for multiview data association,” IEEE Transactions on Robotics , vol. 36, no. 6, pp. 1686–1703, 2020

    2020

  13. [21]

    Fast, accurate and memory-efficient partial permutation synchronization,

    S. Li, Y . Shi, and G. Lerman, “Fast, accurate and memory-efficient partial permutation synchronization,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pp. 15735– 15743, 2022

    2022

  14. [22]

    Factorized graph matching,

    F. Zhou and F. De la Torre, “Factorized graph matching,” IEEE trans- actions on pattern analysis and machine intelligence , vol. 38, no. 9, pp. 1774–1789, 2015

    2015

  15. [23]

    Towards optimizing large-scale multi-graph matching in bioimaging,

    M. Kahl, S. Stricker, L. Hutschenreiter, F. Bernard, C. Rother, and B. Savchynskyy, “Towards optimizing large-scale multi-graph matching in bioimaging,” in Proceedings of the Computer Vision and Pattern Recognition Conference, pp. 11569–11578, 2025

    2025

  16. [24]

    Large-scale multi-view spectral clustering via bipartite graph,

    Y . Li, F. Nie, H. Huang, and J. Huang, “Large-scale multi-view spectral clustering via bipartite graph,” in Proceedings of the AAAI conference on artificial intelligence , vol. 29, 2015

    2015

  17. [25]

    A greedy algorithm for neighborhood overlap-based community detection,

    N. Meghanathan, “A greedy algorithm for neighborhood overlap-based community detection,” Algorithms, vol. 9, no. 1, p. 8, 2016

    2016

  18. [26]

    C ¸ etinkaya-Rundel and J

    M. C ¸ etinkaya-Rundel and J. Hardin, Introduction to Modern Statistics . OpenIntro, 2 ed., 2024. Open-access textbook

    2024

  19. [27]

    Orb: An efficient alternative to sift or surf,

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

    2011

  20. [28]

    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

  21. [29]

    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, pp. 224–236, 2018

    2018

  22. [30]

    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 Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, pp. 4938–4947, 2020

    2020

  23. [31]

    Clus- tergnn: Cluster-based coarse-to-fine graph neural network for efficient feature matching,

    Y . Shi, J.-X. Cai, Y . Shavit, T.-J. Mu, W. Feng, and K. Zhang, “Clus- tergnn: Cluster-based coarse-to-fine graph neural network for efficient feature matching,” in Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , pp. 12517–12526, 2022

    2022

  24. [32]

    Comatcher: Multi-view collaborative feature matching,

    J. Zhang, Z. Xia, M. Dong, S. Shen, L. Yue, and X. Zheng, “Comatcher: Multi-view collaborative feature matching,” in Proceedings of the Com- puter Vision and Pattern Recognition Conference , pp. 21970–21980, 2025

    2025

  25. [33]

    End2end multi-view feature matching with differentiable pose optimization,

    B. Roessle and M. Nießner, “End2end multi-view feature matching with differentiable pose optimization,” in Proceedings of the IEEE/CVF International Conference on Computer Vision , pp. 477–487, 2023

    2023

  26. [34]

    Practical and efficient multi- view matching,

    E. Maset, F. Arrigoni, and A. Fusiello, “Practical and efficient multi- view matching,” in Proceedings of the IEEE International Conference on Computer Vision , pp. 4568–4576, 2017

    2017

  27. [35]

    Unpaired multi-view graph clustering with cross-view structure match- ing,

    Y . Wen, S. Wang, Q. Liao, W. Liang, K. Liang, X. Wan, and X. Liu, “Unpaired multi-view graph clustering with cross-view structure match- ing,” IEEE Transactions on Neural Networks and Learning Systems , 2023

    2023

  28. [36]

    Attention is all you need,

    A. Vaswani, N. Shazeer, N. Parmar, J. Uszkoreit, L. Jones, A. N. Gomez, Ł. Kaiser, and I. Polosukhin, “Attention is all you need,” Advances in neural information processing systems , vol. 30, 2017

    2017

  29. [37]

    An image is worth 16x16 words: Transformers for image recognition at scale,

    A. Dosovitskiy, L. Beyer, A. Kolesnikov, D. Weissenborn, X. Zhai, T. Unterthiner, M. Dehghani, M. Minderer, G. Heigold, S. Gelly, et al., “An image is worth 16x16 words: Transformers for image recognition at scale,” arXiv preprint arXiv:2010.11929 , 2020

    2010 arXiv

  30. [38]

    The graph neural network model,

    F. Scarselli, M. Gori, A. C. Tsoi, M. Hagenbuchner, and G. Monfar- dini, “The graph neural network model,” IEEE transactions on neural networks, vol. 20, no. 1, pp. 61–80, 2008

    2008

  31. [39]

    A comprehensive survey on graph neural networks,

    Z. Wu, S. Pan, F. Chen, G. Long, C. Zhang, and P. S. Yu, “A comprehensive survey on graph neural networks,” IEEE transactions on neural networks and learning systems , vol. 32, no. 1, pp. 4–24, 2020

    2020

  32. [40]

    Graph- based 3d multi-person pose estimation using multi-view images,

    S. Wu, S. Jin, W. Liu, L. Bai, C. Qian, D. Liu, and W. Ouyang, “Graph- based 3d multi-person pose estimation using multi-view images,” in Proceedings of the IEEE/CVF international conference on computer vision, pp. 11148–11157, 2021

    2021

  33. [41]

    Graph neural networks for cross-camera data association,

    E. Luna, J. C. SanMiguel, J. M. Mart ´ınez, and P. Carballeira, “Graph neural networks for cross-camera data association,” IEEE Transactions on Circuits and Systems for Video Technology , vol. 33, no. 2, pp. 589– 601, 2022

    2022

  34. [42]

    Multi-person 3d pose estimation from unlabelled data,

    D. Rodriguez-Criado, P. Bachiller-Burgos, G. V ogiatzis, and L. J. Manso, “Multi-person 3d pose estimation from unlabelled data,” Machine Vision and Applications, vol. 35, no. 3, p. 46, 2024

    2024

  35. [43]

    A noniterative greedy algorithm for multi- frame point correspondence,

    K. Shafique and M. Shah, “A noniterative greedy algorithm for multi- frame point correspondence,” IEEE transactions on pattern analysis and machine intelligence, vol. 27, no. 1, pp. 51–65, 2005

    2005

  36. [44]

    Wide baseline feature matching using the cross-epipolar ordering constraint,

    X. Lu and R. Manduchi, “Wide baseline feature matching using the cross-epipolar ordering constraint,” in Proceedings of the 2004 IEEE Computer Society Conference on Computer Vision and Pattern Recog- nition, 2004. CVPR 2004. , vol. 1, pp. I–I, IEEE, 2004

    2004

  37. [45]

    Multi-camera scene reconstruction via graph cuts,

    V . Kolmogorov and R. Zabih, “Multi-camera scene reconstruction via graph cuts,” in Computer Vision—ECCV 2002: 7th European Conference on Computer Vision Copenhagen, Denmark, May 28–31, 2002 Proceed- ings, Part III 7 , pp. 82–96, Springer, 2002

    2002

  38. [46]

    Multi-view stereo via vol- umetric graph-cuts,

    G. V ogiatzis, P. H. Torr, and R. Cipolla, “Multi-view stereo via vol- umetric graph-cuts,” in 2005 IEEE Computer Society Conference on Computer Vision and Pattern Recognition (CVPR’05) , vol. 2, pp. 391– 398, IEEE, 2005

    2005

  39. [47]

    Solving the multi-way matching problem by permutation synchronization,

    D. Pachauri, R. Kondor, and V . Singh, “Solving the multi-way matching problem by permutation synchronization,” Advances in neural informa- tion processing systems , vol. 26, 2013

    2013

  40. [48]

    Multi-graph matching via affinity optimization with graduated consistency regularization,

    J. Yan, M. Cho, H. Zha, X. Yang, and S. M. Chu, “Multi-graph matching via affinity optimization with graduated consistency regularization,” IEEE transactions on pattern analysis and machine intelligence, vol. 38, no. 6, pp. 1228–1242, 2015

    2015

  41. [49]

    A convex relaxation for multi-graph matching,

    P. Swoboda, A. Mokarian, C. Theobalt, F. Bernard, et al. , “A convex relaxation for multi-graph matching,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pp. 11156– 11165, 2019

    2019

  42. [50]

    Robust bundle adjustment revisited,

    C. Zach, “Robust bundle adjustment revisited,” in Computer Vision– ECCV 2014: 13th European Conference, Zurich, Switzerland, September 6-12, 2014, Proceedings, Part V 13 , pp. 772–787, Springer, 2014

    2014

  43. [51]

    Rkhs-ba: A robust correspondence-free multi-view registration framework with semantic point clouds,

    R. Zhang, J. Song, X. Gao, J. Wu, T. Liu, J. Zhang, R. Eustice, and M. Ghaffari, “Rkhs-ba: A robust correspondence-free multi-view registration framework with semantic point clouds,” arXiv preprint arXiv:2403.01254, 2024

    2024 arXiv

  44. [52]

    On spectral clustering: Analysis and an algorithm,

    A. Ng, M. Jordan, and Y . Weiss, “On spectral clustering: Analysis and an algorithm,” Advances in neural information processing systems , vol. 14, 2001

    2001

  45. [53]

    A clustering method for efficient segmentation of 3d laser data,

    K. Klasing, D. Wollherr, and M. Buss, “A clustering method for efficient segmentation of 3d laser data,” in 2008 IEEE international conference on robotics and automation , pp. 4043–4048, IEEE, 2008

    2008

  46. [54]

    Realtime multi-person 2d pose estimation using part affinity fields,

    Z. Cao, T. Simon, S.-E. Wei, and Y . Sheikh, “Realtime multi-person 2d pose estimation using part affinity fields,” in Proceedings of the IEEE conference on computer vision and pattern recognition , pp. 7291–7299, 2017

    2017

  47. [55]

    Hennig, M

    C. Hennig, M. Meila, F. Murtagh, and R. Rocci, Handbook of cluster analysis. CRC press, 2015

    2015

  48. [56]

    3d pictorial structures for multiple human pose estimation,

    V . Belagiannis, S. Amin, M. Andriluka, B. Schiele, N. Navab, and S. Ilic, “3d pictorial structures for multiple human pose estimation,” in Proceedings of the IEEE conference on computer vision and pattern recognition, pp. 1669–1676, 2014

    2014

  49. [57]

    Panoptic studio: A massively multiview system for social motion capture,

    H. Joo, H. Liu, L. Tan, L. Gui, B. Nabbe, I. Matthews, T. Kanade, S. Nobuhara, and Y . Sheikh, “Panoptic studio: A massively multiview system for social motion capture,” in Proceedings of the IEEE interna- tional conference on computer vision , pp. 3334–3342, 2015

    2015

  50. [58]

    Easyvis2: A real time multi-view 3d visualization for laparoscopic surgery training enhanced by a deep neural network yolov8-pose,

    Y .-H. Sun, G. Shen, J. Chen, J. Fernandes, H. Jiang, and Y . H. Hu, “Easyvis2: A real time multi-view 3d visualization for laparoscopic surgery training enhanced by a deep neural network yolov8-pose,” arXiv preprint arXiv:2412.16742, 2024

    2024 arXiv

  51. [59]

    Easyvis: a real-time 3d visualization software system for laparoscopic surgery box trainer,

    Y .-H. Sun, J. Ke, J. Fernandes, J. Chen, H. Jiang, and Y . H. Hu, “Easyvis: a real-time 3d visualization software system for laparoscopic surgery box trainer,” Updates in Surgery , pp. 1–16, 2025

    2025

  52. [60]

    Openmvg: Open mul- tiple view geometry,

    P. Moulon, P. Monasse, R. Perrot, and R. Marlet, “Openmvg: Open mul- tiple view geometry,” in Reproducible Research in Pattern Recognition: First International Workshop, RRPR 2016, Canc ´un, Mexico, December 4, 2016, Revised Selected Papers 1 , pp. 60–74, Springer, 2017

    2016

Pith tools

Reviewed August 6, 2026 · model on record in the stance chip above.