Pith. sign in

REVIEW 4 major objections 6 minor 32 references

GMFlow: Global Motion-Guided Recurrent Flow for 6D Object Pose Estimation

T0 review · 4 major / 6 minor · reviewed 2026-08-12 · deepseek-v4-flash

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

Pith's one-line read The paper claims that a global motion-guided recurrent flow, which transfers motion from visible object parts to occluded ones, achieves the best reported 6D pose refinement accuracy on LM-O and YCB-V among compared methods.

desk verdict Solid incremental refinement paper whose empirical gains are plausible, but the occlusion-transfer mechanism is overclaimed and it needs to own SCFlow's shape-constraint lineage. read the letter →

arxiv 2411.17174 v1 pith:XC7JCBUF submitted 2024-11-26 cs.CV

Xin Liu , Shibei Xue , Dezong Zhao , Shan Ma , Min Jiang This is my paper · ORCID
classification cs.CV
keywords 6Dobjectposeestimationrefinementrecurrentflowglobalmotionaggregationocclusionhandlingrender-and-comparelinearattentionrigidbodyconstraints
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

This paper tries to establish that flow-based 6D pose refinement can survive occlusion and partial cropping if the flow estimator stops trusting only local pixel correspondences and instead seeks a global explanation of object motion. The proposed GMFlow method draws on the rigid-body property that all points of an object move together, so motion observed on visible parts can be extrapolated to invisible parts. It encodes that idea with a global motion capture module, a linear attention step that pools object-wide context into a vector and multiplies it into local motion features, plus a pose-to-flow projection that feeds shape-constrained flow back into the recurrent iterations. On the occluded-object benchmarks LM-O and YCB-V, the paper reports the best average recall among compared refinement methods (67.0 and 74.2 ADD-0.1d) at 13–18 ms per image, suggesting the gains do not come at prohibitive runtime cost. The claim is that global context plus rigid-shape constraints, not more iterations or heavier rendering, is what makes flow-based refinement work under occlusion.

What carries the argument

The Global Motion Capture (GMC) module is the central object: it computes a global context vector from context features via a learned linear attention query $q = \theta(F_c) \cdot w_a / \sqrt{D}$, normalizes $q$ over the spatial dimension, forms $g = \sum_i q_i C_i$, and produces global motion features $F_g = \tau(\alpha \cdot (C + \phi(M \odot g)))$, where $\odot$ is element-wise multiplication. This module is what transfers motion information from visible to invisible regions. The second load-bearing mechanism is the pose-to-flow projection: after each GRU update predicts a pose correction, the 3D model points are re-projected under both the initial and corrected poses, and the displacement field $f_k$ is used to query the correlation volume in the next iteration, enforcing rigid-body consistency on the flow.

What would settle it

Train GMFlow and an ablation without the global motion capture module on the same data, then evaluate on a synthetic test set where each object is artificially occluded by randomly placed blocks at occlusion ratios from 0.5 to 0.95; if the ADD-0.1d gap between the full model and the ablation does not grow with occlusion ratio, or if per-pixel flow error on occluded pixels is not reduced by the global features, the paper's central mechanism is not doing the claimed work.

Watch

Extended reading notes

Core claim

The central claim is that local motion features alone are ambiguous under occlusion, and that the ambiguity can be resolved by a global motion explanation computed from the object's overall appearance. The Global Motion Capture module builds a global context vector $g = \sum_{i=1}^{N} q_i C_i$ from context features $C$, where the attention weights $q$ come from a learned linear projection followed by normalization, and then gates the local motion features $M$ by this vector to produce global motion features. The recurrent pose update operator then closes the loop with the object's 3D shape: instead of using the predicted flow directly as the next query, it projects 3D model points under the corrected pose and the initial pose and computes the pose-induced flow $f_k = u_{k,i} - u_{0,i}$, which becomes the query for the next correlation lookup. This shape-constrained flow ties every iteration to rigid-body geometry. The paper reports that with four or eight iterations, GMFlow outperforms prior flow-based refinement methods on both LM-O and YCB-V under ADD-based and BOP metrics.

Load-bearing premise

The load-bearing premise is that, because the object is rigid, the 2D motion between rendered and real images is similar and uniform enough across the object that global context learned from visible parts correctly fills in motion for invisible parts, and that this transfer generalizes beyond the training distribution.

Editorial extensions

If this is right

  • If the reported results hold, robotic manipulators can refine 6D poses with a single RGB camera under heavy occlusion, because global motion guidance removes the need for visible correspondences on every object part.
  • The method reaches near-final accuracy in about two iterations, so pose refinement can run at interactive rates (13–18 ms per object), which is within the range needed for closed-loop robot control.
  • Because the paper shows refinement on top of both PoseCNN and WDR initializations, the module could be attached to any coarse pose estimator rather than being tied to one detector.
  • The pose-to-flow constraint implies that flow estimates stay consistent with rigid-body geometry every iteration, so the same loop should transfer to other render-and-compare tasks on rigid objects, such as tracking or grasping.

Reading between the lines

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

  • The paper's assumption that object motion is uniform in 2D is only approximate: perspective projection makes pixel displacement depth-dependent, so the learned global transfer may not hold for unseen camera intrinsics or unusually shaped objects, and the paper does not stress-test this.
  • A natural extension the paper does not explore is explicitly supervising the global context vector with the object's visible mask or 3D model, which could make the transfer more interpretable and robust to large occluders.
  • The reported runtime counts a single-object crop; in multi-object scenes the cost scales with the number of instances, so the 13–18 ms figure should not be read as a full-scene latency.
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 / 6 minor

Summary. The manuscript proposes GMFlow, a recurrent render-and-compare pose refinement method that augments RAFT-style flow estimation with a Global Motion Capture (GMC) module based on linear attention and a pose-to-flow projection constrained by the object's 3D shape. The pipeline iteratively refines a 6D pose by estimating flow between a rendered image and the target image, aggregating global contextual information from the rendered image, and using the updated pose to compute a shape-constrained flow for the next iteration. The method is evaluated on LM-O and YCB-V with ADD/ADD-S and BOP metrics, reporting state-of-the-art average recall (67.0 on LM-O and 74.2 on YCB-V at ADD-0.1d) and competitive runtimes of 13-18 ms per image.

Significance. If the reported results are reproducible, GMFlow represents a practically useful advance in RGB-only 6D pose refinement: it requires only a single rendering pass, converges in a few iterations, and consistently outperforms PFA and SCFlow in the reported tables, including under varying occlusion levels and initial pose noise. The ablation study is clearly structured and the training/evaluation protocol is standard for the field. The main weaknesses are the lack of code release, the absence of error bars or significance tests for the performance claims, and a mechanistic gap between the GMC module's architecture and the occlusion-transfer narrative used to motivate it.

major comments (4)
  1. [Section III-B, Eqs. (1)-(3)] The GMC module computes its global context vector g from θ(Fc), where Fc is the context feature of the rendered image, and then applies g as a channel-wise multiplicative gain to the local motion feature M everywhere. This is a global gain, not a spatial propagation of motion from visible to invisible pixels. The claim in Section I that the method can "transfer motion information from visible parts to invisible parts" is therefore not established by the architecture. The ablation in Table III (row B0) removes the entire module, so the reported 2.1-2.3 point drop could reflect added capacity or an extra input stream rather than any occlusion-specific mechanism. Please provide a direct evaluation of flow or pose errors on occluded versus non-occluded pixels, and an ablation that replaces the rendered-context-derived g with a target-image-context or combined-context g, to test whether the module specifically handles occlusion.
  2. [Section III-C, Eqs. (4)-(5)] The sentence "Unlike traditional methods [8], [10], [25], we do not directly use the predicted flow finter as the query input for the next iteration" is inaccurate regarding SCFlow [10], whose "shape-constraint recurrent flow" already projects 3D model points with the current pose to compute a 2D flow for querying the correlation volume. Because the abstract and related work present shape constraints as a new contribution, the manuscript must clearly differentiate GMFlow's pose-to-flow step from SCFlow's, reposition the shape constraint as a shared component, and provide an ablation that isolates what is genuinely new in the pose-to-flow design.
  3. [Section IV-B, Tables I and II] The reported advantages over SCFlow are small in several columns (e.g., +0.6 percentage points on LM-O ADD-0.1d, +0.007 average BOP on YCB-V). The paper reports a single run per configuration without variance, confidence intervals, or significance tests. Since pose refinement results can be sensitive to initialization and training randomness, please report results over multiple seeds with error bars, or use a per-object paired statistical test, to substantiate the claim that GMFlow "outperforms existing techniques."
  4. [Section I] The motivating statement that "the motion of different points on an object is similar and uniform" under rigid-body motion, and that this applies to the 2D motion between rendered and real images, is not correct for perspective projection: the projected 2D flow depends on depth and is generally non-uniform across the object. If this is intended only as high-level intuition, it should be phrased in terms of 3D rigidity (i.e., a single rigid transformation). If it is used to justify the global transfer mechanism, the paper should provide evidence that the learned global aggregation can explain depth-dependent 2D motion, for instance by visualizing or analyzing flow fields across depth.
minor comments (6)
  1. [Table IV] The heading "Timeing" should read "Timing."
  2. [Section III-C] There are typographical errors: "methodsn" should be "methods" and "iteratio" should be "iteration."
  3. [Section IV-A] The LM-O dataset is cited as reference [16], which is Cho et al. on neural machine translation rather than the original LM-O dataset source; please correct the citation.
  4. [Section IV-C] The phrase "grid diameter" in the ablation discussion should be "object diameter" (or "model diameter").
  5. [Figure 1] The captions "Cat's rendered image" and "Can's rendered image" should clarify that these are renderings from the initial pose of the corresponding object model, not the real occluded images.
  6. [Equations (1)-(2)] The normalization that makes the attention query vector q sum to unity is not explicitly specified; please state whether it is a softmax or another normalization.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: all reported results are held-out benchmark evaluations of a trained model, with no fitted input renamed as prediction and no load-bearing self-citation chain.

full rationale

GMFlow is a trained pose-refinement network evaluated on held-out test splits of LM-O and YCB-V. The reported accuracies (Tables I and II) are benchmark measurements, not quantities defined in terms of the method's own outputs. The loss function (Eq. 6) combines pose and flow objectives with fixed hyperparameters (gamma = 0.8, k = 0.1), and the ground-truth flow is computed from the initial and actual poses; this is a standard supervised training setup, not a self-referential definition. The GMC module (Eqs. 1-3) is a learned attention mechanism whose behavior on occluded pixels is an empirical question; whether the context vector computed from the rendered image can actually encode occlusion is a correctness or generalization concern, not a circularity. The one reference co-authored by a paper author, [20], is a minor related-work citation for high-resolution keypoint-based pose estimation and is not used to justify the central claim. The ablation study (Table III) compares ablations of the same trained framework, which is standard practice and does not reduce the headline results to fitted constants. No equation in the paper defines its own prediction, no fitted parameter is renamed as a predicted result, and no uniqueness theorem or self-citation chain is invoked to force the method's design. Accordingly, the circularity score is 0.

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

This is an empirical deep-learning paper; the main free component is the trained network itself. There are no new physical entities. The method's design rests on the rigid-body motion transfer premise and the render-and-compare assumption, both standard for the field.

free parameters (6)
  • loss balancing factor k = 0.1
    Hand-chosen in Section III-D, Eq. (6); weights pose vs flow losses, affects training and final accuracy.
  • flow loss decay gamma = 0.8
    Hand-chosen in Section III-D, Eq. (6); exponentially weights iteration losses, affecting convergence behavior.
  • number of iterations N = 4 and 8
    Section III-D; reporting both, more iterations improve accuracy but add runtime, so the choice affects the accuracy-efficiency tradeoff.
  • global modulation scalar alpha = learned
    Learnable scalar in Eq. (3) modulating the influence of global information; trained on benchmark data.
  • global attention query vector w_a = learned
    Learnable parameter in Eq. (1); the GMC module's attention weights, trained on benchmark data.
  • all network weights (CNN encoders, GRU, projection functions) = trained end-to-end on LM-O/YCB-V
    The central accuracy claim is empirical, and the entire network capacity is the main fitted component; the paper provides training hyperparameters but no pre-trained weights.
assumptions (4)
  • domain assumption The motion of different points on an object is similar and uniform under the rigid body assumption, and this applies to 2D flow between rendered and real images.
    Stated in Section I; underpins the GMC design, though 2D projected motion is depth-dependent, so the assumption is only approximate.
  • domain assumption The render-and-compare setup assumes an initial pose P0 close enough to the true pose that a meaningful rendered image is available for comparison.
    Section III; the method refines an initial pose; without a reasonable initialization the flow range exceeds the correlation search radius.
  • domain assumption Camera intrinsics K and the full 3D object model M are known and accurate.
    Eq. (4) requires K and the object model; the paper uses Pytorch3D rendering and standard datasets; inaccurate intrinsics would break pose-to-flow projection.
  • domain assumption Supervised training uses ground-truth poses and rendered images; the network is trained and evaluated on the same benchmarks it tunes hyperparameters on.
    Losses in Section III-D use GT poses; no cross-dataset generalization is demonstrated, so in-distribution benchmarking is the evidence base.

how reviews work

0 comments
Cite this review

Pith. "Pith review of GMFlow: Global Motion-Guided Recurrent Flow for 6D Object Pose Estimation." pith.science (2026). https://pith.science/paper/XC7JCBUF

@misc{pith2026241117174,
  author       = {Pith},
  title        = {Pith review of: GMFlow: Global Motion-Guided Recurrent Flow for 6D Object Pose Estimation},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/XC7JCBUF}},
  note         = {Machine review of arXiv:2411.17174}
}
read the original abstract

6D object pose estimation is crucial for robotic perception and precise manipulation. Occlusion and incomplete object visibility are common challenges in this task, but existing pose refinement methods often struggle to handle these issues effectively. To tackle this problem, we propose a global motion-guided recurrent flow estimation method called GMFlow for pose estimation. GMFlow overcomes local ambiguities caused by occlusion or missing parts by seeking global explanations. We leverage the object's structural information to extend the motion of visible parts of the rigid body to its invisible regions. Specifically, we capture global contextual information through a linear attention mechanism and guide local motion information to generate global motion estimates. Furthermore, we introduce object shape constraints in the flow iteration process, making flow estimation suitable for pose estimation scenarios. Experiments on the LM-O and YCB-V datasets demonstrate that our method outperforms existing techniques in accuracy while maintaining competitive computational efficiency.

Figures

Figures reproduced from arXiv: 2411.17174 by the authors.

Figure 1
Figure 1. Challenges of flow-based method in 6D pose es￾timation task. Flow is commonly used to estimate the pixel motion vector field between two frames in an image sequence. Examples are shown in (a) and (b) [1]. The objects in the images are not necessarily rigid bodies, and their changes are typically limited in magnitude. However, pose estimation tasks differ in this regard. Although rendered images, such as (d) and (f),… view at source ↗
Figure 2
Figure 2. Overview of the proposed method. features from the target and rendered images are first encoded and then associated through local motion capture and global motion aggregation. During the pose refinement stage, flow, correlation, aggregated information, and the hidden state h k−1 are simultaneously input into the GRU [16]. The GRU initially predicts an intermediate flow and subsequently generates pose correction. By … view at source ↗
Figure 3
Figure 3. The structure diagram of the global motion capture [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗
Figures from the paper (4 more)
Figure 4
Figure 4. Figure 4: Intermediate Flow Comparison. Our method demonstrates superior handling of occlusions and better utilization of contextual information in early iterations. it with the flow generated by the flow regressor. To quantify prediction errors, we use the L1 loss. Considering …
Figure 5
Figure 5. Figure 5: Qualitative results. From left to right: target image, intermediate flow with reconstruction, pose-induced flow with reconstruction, and comparison before and after our method. Three rows illustrate cases where the target object is complete, occluded, and partially cro…
Figure 6
Figure 6. Figure 6: Results under different degrees of occlusion. [PITH_FULL_IMAGE:figures/full_fig_p007_6.png]
Figure 7
Figure 7. Figure 7: Comparison of robustness and efficiency. [PITH_FULL_IMAGE:figures/full_fig_p007_7.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

32 extracted references · 26 canonical work pages

  1. [10]

    Shape-constraint recurrent flow for 6D object pose estimation,

    Y . Hai, R. Song, J. Li, and Y . Hu, “Shape-constraint recurrent flow for 6D object pose estimation,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , 2023, pp. 4831–4840

    work page 2023

  2. [8]

    Perspective flow aggregation for data-limited 6d object pose estimation,

    Y . Hu, P. Fua, and M. Salzmann, “Perspective flow aggregation for data-limited 6d object pose estimation,” in European Conference on Computer Vision. Springer, 2022, pp. 89–106

    work page 2022

  3. [25]

    Rnnpose: Recurrent 6- Dof object pose refinement with robust correspondence field estimation and pose optimization,

    Y . Xu, K.-Y . Lin, G. Zhang, X. Wang, and H. Li, “Rnnpose: Recurrent 6- Dof object pose refinement with robust correspondence field estimation and pose optimization,” in Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , 2022, pp. 14 880–14 890

    work page 2022

  4. [1]

    Slow flow: Exploiting high-speed cameras for accurate and diverse optical flow reference data,

    J. Janai, F. Guney, J. Wulff, M. J. Black, and A. Geiger, “Slow flow: Exploiting high-speed cameras for accurate and diverse optical flow reference data,” in Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition , 2017, pp. 3597–3607

    work page 2017

  5. [2]

    Learning analysis-by-synthesis for 6D pose estimation in RGB-D images,

    A. Krull, E. Brachmann, F. Michel, M. Y . Yang, S. Gumhold, and C. Rother, “Learning analysis-by-synthesis for 6D pose estimation in RGB-D images,” in Proceedings of the IEEE international conference on computer vision , 2015, pp. 954–962

    work page 2015

  6. [3]

    Posecnn: a convolu- tional neural network for 6D object pose estimation in cluttered scenes,

    Y . Xiang, T. Schmidt, V . Narayanan, and D. Fox, “Posecnn: a convolu- tional neural network for 6D object pose estimation in cluttered scenes,” Robotics: Science and Systems (RSS), , 2018

    work page 2018

  7. [4]

    Bop challenge 2022 on detection, segmentation and pose estimation of specific rigid objects,

    M. Sundermeyer, T. Hoda ˇn, Y . Labbe, G. Wang, E. Brachmann, B. Drost, C. Rother, and J. Matas, “Bop challenge 2022 on detection, segmentation and pose estimation of specific rigid objects,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , 2023, pp. 2785–2794

    work page 2022

  8. [5]

    A method for registration of 3-D shapes,

    P. Besl and N. D. McKay, “A method for registration of 3-D shapes,” IEEE Transactions on Pattern Analysis & Machine Intelligence, vol. 14, no. 02, pp. 239–256, 1992

    work page 1992

Show all 32 references
  1. [6]

    Deepim: Deep iterative matching for 6d pose estimation,

    Y . Li, G. Wang, X. Ji, Y . Xiang, and D. Fox, “Deepim: Deep iterative matching for 6d pose estimation,” in Proceedings of the European Conference on Computer Vision (ECCV) , 2018, pp. 683–698

    2018

  2. [7]

    Cosypose: Consistent multi-view multi-object 6D pose estimation,

    Y . Labb ´e, J. Carpentier, M. Aubry, and J. Sivic, “Cosypose: Consistent multi-view multi-object 6D pose estimation,” in Computer Vision–ECCV 2020: 16th European Conference, Glasgow, UK, August 23–28, 2020, Proceedings, Part XVII 16 . Springer, 2020, pp. 574–591

    2020

  3. [9]

    Raft: Recurrent all-pairs field transforms for op- tical flow,

    Z. Teed and J. Deng, “Raft: Recurrent all-pairs field transforms for op- tical flow,” inComputer Vision–ECCV 2020: 16th European Conference, Glasgow, UK, August 23–28, 2020, Proceedings, Part II 16 . Springer, 2020, pp. 402–419

    2020

  4. [11]

    Accelerating 3D deep learning with Pytorch3D,

    N. Ravi, J. Reizenstein, D. Novotny, T. Gordon, W.-Y . Lo, J. Johnson, and G. Gkioxari, “Accelerating 3D deep learning with Pytorch3D,” arXiv:2007.08501, 2020

    2007 arXiv

  5. [12]

    Using relaxation to find a puppet,

    G. Hinton, “Using relaxation to find a puppet,” in Proceedings of the 2nd Summer Conference on Artificial Intelligence and Simulation of Behaviour, 1976, pp. 148–157

    1976

  6. [13]

    Attention is all you need,

    A. Vaswani, “Attention is all you need,” Advances in Neural Information Processing Systems, 2017

    2017

  7. [14]

    Learning to esti- mate hidden motions with global motion aggregation,

    S. Jiang, D. Campbell, Y . Lu, H. Li, and R. Hartley, “Learning to esti- mate hidden motions with global motion aggregation,” in Proceedings of the IEEE/CVF international conference on computer vision , 2021, pp. 9772–9781

    2021

  8. [15]

    Swiftformer: Efficient additive attention for transformer-based real-time mobile vision applications,

    A. Shaker, M. Maaz, H. Rasheed, S. Khan, M.-H. Yang, and F. S. Khan, “Swiftformer: Efficient additive attention for transformer-based real-time mobile vision applications,” in Proceedings of the IEEE/CVF International Conference on Computer Vision, 2023, pp. 17 425–17 436

    2023

  9. [16]

    On the properties of neural machine translation: Encoder-decoder approaches,

    K. Cho, B. Van Merri ¨enboer, D. Bahdanau, and Y . Bengio, “On the properties of neural machine translation: Encoder-decoder approaches,” arXiv preprint arXiv:1409.1259 , 2014

    2014 arXiv

  10. [17]

    Gdr-net: Geometry- guided direct regression network for monocular 6D object pose estima- tion,

    G. Wang, F. Manhardt, F. Tombari, and X. Ji, “Gdr-net: Geometry- guided direct regression network for monocular 6D object pose estima- tion,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2021, pp. 16 611–16 621

    2021

  11. [18]

    Poet: Pose estimation transformer for single-view, multi-object 6D pose estimation,

    T. G. Jantos, M. A. Hamdad, W. Granig, S. Weiss, and J. Steinbrener, “Poet: Pose estimation transformer for single-view, multi-object 6D pose estimation,” in Conference on Robot Learning. PMLR, 2023, pp. 1060– 1070

    2023

  12. [19]

    Pvnet: Pixel- wise voting network for 6DoF pose estimation,

    S. Peng, Y . Liu, Q. Huang, X. Zhou, and H. Bao, “Pvnet: Pixel- wise voting network for 6DoF pose estimation,” in Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, 2019, pp. 4561–4570

    2019

  13. [20]

    High-resolution representation ob- ject pose estimation from monocular images,

    Q. Guan, W. Li, S. Xue, and D. Li, “High-resolution representation ob- ject pose estimation from monocular images,” in 2021 China Automation Congress (CAC). IEEE, 2021, pp. 980–984

    2021

  14. [21]

    Dpodv2: Dense correspondence- based 6 DoF pose estimation,

    I. Shugurov, S. Zakharov, and S. Ilic, “Dpodv2: Dense correspondence- based 6 DoF pose estimation,” IEEE transactions on pattern analysis and machine intelligence , vol. 44, no. 11, pp. 7417–7435, 2021

    2021

  15. [22]

    Surfemb: Dense and continuous correspondence distributions for object pose estimation with learnt surface embeddings,

    R. L. Haugaard and A. G. Buch, “Surfemb: Dense and continuous correspondence distributions for object pose estimation with learnt surface embeddings,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , 2022, pp. 6749–6758

    2022

  16. [23]

    Epnp: An accurate o(n) solution to the PnP problem,

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

    2009

  17. [24]

    Epro-pnp: Generalized end-to-end probabilistic perspective-n-points for monocular object pose estimation,

    H. Chen, P. Wang, F. Wang, W. Tian, L. Xiong, and H. Li, “Epro-pnp: Generalized end-to-end probabilistic perspective-n-points for monocular object pose estimation,” in Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , 2022, pp. 2781–2790

    2022

  18. [26]

    Repose: Fast 6D object pose refinement via deep texture rendering,

    S. Iwase, X. Liu, R. Khirodkar, R. Yokota, and K. M. Kitani, “Repose: Fast 6D object pose refinement via deep texture rendering,” in Proceed- ings of the IEEE/CVF International Conference on Computer Vision , 2021, pp. 3303–3312

    2021

  19. [27]

    Super-convergence: Very fast training of neural networks using large learning rates,

    L. N. Smith and N. Topin, “Super-convergence: Very fast training of neural networks using large learning rates,” in Artificial intelligence and machine learning for multi-domain operations applications , vol. 11006. SPIE, 2019, pp. 369–386

    2019

  20. [28]

    So-pose: Exploiting self-occlusion for direct 6D pose estimation,

    Y . Di, F. Manhardt, G. Wang, X. Ji, N. Navab, and F. Tombari, “So-pose: Exploiting self-occlusion for direct 6D pose estimation,” in Proceedings of the IEEE/CVF International Conference on Computer Vision , 2021, pp. 12 396–12 405

    2021

  21. [29]

    Coupled iterative refinement for 6D multi-object pose estimation,

    L. Lipson, Z. Teed, A. Goyal, and J. Deng, “Coupled iterative refinement for 6D multi-object pose estimation,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , 2022, pp. 6728–6737

    2022

  22. [30]

    Sc6d: Symmetry-agnostic and correspondence-free 6D object pose estimation,

    D. Cai, J. Heikkil ¨a, and E. Rahtu, “Sc6d: Symmetry-agnostic and correspondence-free 6D object pose estimation,” in 2022 International Conference on 3D Vision (3DV) . IEEE, 2022, pp. 536–546

    2022

  23. [31]

    Bop challenge 2020 on 6D object localization,

    T. Hoda ˇn, M. Sundermeyer, B. Drost, Y . Labb ´e, E. Brachmann, F. Michel, C. Rother, and J. Matas, “Bop challenge 2020 on 6D object localization,” in Computer Vision–ECCV 2020 Workshops: Glasgow, UK, August 23–28, 2020, Proceedings, Part II 16 . Springer, 2020, pp. 577–594

    2020

  24. [32]

    Wide-depth- range 6D object pose estimation in space,

    Y . Hu, S. Speierer, W. Jakob, P. Fua, and M. Salzmann, “Wide-depth- range 6D object pose estimation in space,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , 2021, pp. 15 870–15 879

    2021

Pith tools

Reviewed August 12, 2026 · model on record in the stance chip above.