Pith. sign in

REVIEW 3 major objections 6 minor 69 references

Diff9D: Diffusion-Based Domain-Generalized Category-Level 9-DoF Object Pose Estimation

T0 review · 3 major / 6 minor · reviewed 2026-08-09 · deepseek-v4-flash

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

Pith's one-line read A diffusion model that treats 9-DoF pose as a vector to denoise sets new standards in synthetic-to-real category-level pose estimation, with no shape priors and only 3 denoising steps.

desk verdict Useful synthetic-only pose diffusion with a real parameterization risk: rotation is diffused as a flat vector, and the paper never checks whether denoised outputs are near-orthogonal. read the letter →

arxiv 2502.02525 v1 pith:N5D5H7LF submitted 2025-02-04 cs.CV cs.RO

Jian Liu , Wei Sun , Hui Yang , Pengchao Deng , Chongpei Liu , Nicu Sebe , Hossein Rahmani , Ajmal Mian This is my paper · ORCID
classification cs.CVcs.RO
keywords category-levelobjectposeestimationdiffusionmodeldomaingeneralizationroboticgrasping9-DoFsynthetic-to-realDDIM
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

The paper claims that category-level 9-DoF object pose and size estimation can be solved from rendered synthetic data alone, if the task is reframed as a generative denoising problem rather than direct regression. It proposes Diff9D, a denoising diffusion model that starts from Gaussian noise and, conditioned on RGB-D observations, denoises a 15-dimensional pose vector (translation, rotation, and size). With the DDIM scheduler, reverse diffusion needs only three steps, giving near real-time performance at 17.2 FPS. Trained exclusively on the CAMERA25 synthetic dataset and deployed on REAL275, Wild6D, and a real robot, the method reports 69.2% 3D50 and 45.2% 5°5cm mAP on REAL275, 76.48% 3D50 on Wild6D, and 80.8% average grasping success, outperforming prior domain-generalization methods and some methods trained on real labeled data. The central claim is that diffusion's sampling process can absorb the synthetic-to-real domain gap without any 3D shape priors or real-world labels.

What carries the argument

The central mechanism is the diffusion process applied to the pose vector itself. The forward process adds Gaussian noise to a 15-dimensional pose vector (translation, rotation matrix entries, and size) according to a variance schedule, reaching a standard Gaussian at the final step; the reverse process, learned by a transformer-based denoiser, predicts the noise conditioned on multimodal features (RGB global features from ResNet18, point-cloud global features from PointNet, time-step embedding, and shape/NOCS shape features from a shape estimator-encoder). The DDIM scheduler then performs deterministic or near-deterministic denoising in as few as 3 steps, instead of the usual 1000. The key modeling choice is that the 3×3 rotation matrix is treated as an ordinary Euclidean vector during noising and denoising, with no projection onto SO(3) or manifold-aware parameterization.

What would settle it

On a real-world test set, compute for each predicted rotation matrix the orthogonality error $\|R^\top R - I\|_F$ and the percentage of outputs whose determinant deviates from 1. If a substantial fraction of high-scoring predictions are far from orthogonal, the Euclidean diffusion is not preserving rotation structure, and the reported rotation accuracy is likely optimistic.

Watch

Extended reading notes

Core claim

Diff9D redefines category-level 9-DoF object pose estimation as a reverse diffusion process: given a noisy pose vector sampled from a Gaussian, the model iteratively denoises it into the true 3D translation, 3×3 rotation matrix, and 3D size, conditioned on features extracted from the observed RGB image, point cloud, time step, and a predicted shape/NOCS shape. The network is trained end-to-end on synthetic data with a loss that combines pose-noise prediction with Chamfer and Smooth-L1 shape losses. Using the DDIM scheduler, the reverse process is run for as few as 3 steps, yielding 17.2 FPS. Because the pose vector is sparse (15 values), the paper argues that far fewer steps are needed than in image diffusion, and that the sampling along the Markov chain broadens the synthetic pose distribution, reducing the domain gap to real scenes. The model does not require any 3D shape priors at training or inference, and the authors claim state-of-the-art domain generalization on REAL275, Wild6D, and in real robotic grasping experiments.

Load-bearing premise

The model treats the 3×3 rotation matrix as an ordinary Euclidean vector when adding and removing noise, without projecting onto the space of valid rotations, and this is assumed to cause no loss in pose accuracy.

Editorial extensions

If this is right

  • Synthetic-only training removes the need for real-world pose annotations, which are costly and labor-intensive to collect.
  • Since no 3D shape priors are used, there is no need to build CAD model libraries per category, simplifying deployment to new object categories.
  • Three-step DDIM denoising makes the method fast enough (17.2 FPS) for real-time robotic manipulation, not just offline evaluation.
  • The same diffusion formulation extends to unseen object instances within a category and to novel scenes, as demonstrated on Wild6D and four robotic grasping scenes.

Reading between the lines

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

  • The paper never measures how close the denoised rotation matrices are to valid rotations. A natural extension would be to compute the orthogonality error (||R^T R − I||) on REAL275 outputs; if errors are small, the Euclidean treatment is benign; if large, the n°mcm metrics may mask invalid rotations.
  • The claim that diffusion makes the pose distribution more uniform is testable: one could estimate the empirical pose distribution of synthetic samples vs. the diffusion-sampled distribution and measure a distributional distance (e.g., Wasserstein) to confirm the broadening effect.
  • The method suggests a general recipe for other geometric regression tasks (e.g., 6D tracking, hand pose, camera pose): treat the target as a sparse vector and denoise it under multimodal conditions. The 3-step convergence indicates the denoiser may be learning an almost deterministic mapping, so a direct comparison with a residual regression network under identical conditioning would isolate what
  • Because the rotation matrix is diffused in Euclidean space, a variant using quaternions or axis-angle plus projection to SO(3) might improve rotation accuracy or require even fewer steps; this is a concrete testable variant the paper does not explore.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

3 major / 6 minor

Summary. The paper proposes Diff9D, a diffusion-based method for category-level 9-DoF object pose and size estimation trained exclusively on rendered synthetic data (CAMERA25) and evaluated on real-world datasets (REAL275, Wild6D) as well as on a real robotic grasping platform. The pose is represented as a 15-dimensional vector (translation, size, and the 3x3 rotation matrix); a forward diffusion process adds Gaussian noise to this vector, and a transformer-based denoiser predicts the added noise conditioned on RGB, point cloud, time step, and learned shape/NOCS features. DDIM sampling allows reverse diffusion in as few as 3 steps, yielding 17.2 FPS. The paper reports state-of-the-art domain generalization results (e.g., 69.2% 3D50 and 45.2% 5°5cm mAP on REAL275 with a synthetic-retrained Mask R-CNN, 76.48% 3D50 on Wild6D, and 80.8% average grasping success), along with extensive ablations of the condition extraction module, the denoiser architecture, and the number of diffusion steps.

Significance. If the reported numbers are valid, Diff9D offers a practical synthetic-only training recipe for category-level 9-DoF pose estimation, eliminating the need for real-world annotations and 3D shape priors. The lightweight denoiser and 3-step DDIM make it relevant for real-time robotic manipulation, and the inclusion of a physical robot evaluation strengthens the application story. The paper is commendable for its thorough ablation coverage. The central claim, however, rests on an empirical assumption about the validity of diffusing rotation matrices as flat Euclidean vectors, and on hyperparameters selected using the same REAL275 benchmark that is later used for the state-of-the-art comparison; both issues need to be resolved before the SOTA claim can be regarded as established.

major comments (3)
  1. [Sec. 3.1 (Eqs. 4, 5, 10); Sec. 3.4 (Eq. 15)] The pose diffusion process treats the 3x3 rotation matrix as a flat 9-dimensional Euclidean vector. The forward process adds Gaussian noise to the rotation-matrix entries (Eqs. 4-5), the DDIM reverse process runs in R^15 (Eq. 10), and the denoising loss is an unconstrained L2 loss on the raw matrix entries (Eq. 15). Since x_T ~ N(0,I) has no rotation structure and no projection to SO(3) is applied, the 9-dimensional output of the network may be non-orthogonal. The n°mcm metrics in Tabs. 1-3 compute rotation error from this raw matrix; if the output is not close to a proper rotation, those numbers are not well-defined. The paper does not report any orthogonality-error statistic (e.g., ||R^T R - I||_F) or any SO(3) post-processing. Please add an orthogonality-error analysis and, if necessary, apply an orthogonal projection before evaluation or adopt a manifold-aware rotation parameterization; the main conclusions should be re-evaluated accordingly.
  2. [Sec. 5.2; Tabs. 7-8; Tab. 1] Several key hyperparameters are selected on the REAL275 test set that is then used for the state-of-the-art comparison: S=3 (Tab. 8), Tb=7 (Tab. 7), and the number of attention heads m=16 (Sec. 5.2). Because REAL275 is also the evaluation benchmark in Tab. 1, this constitutes a form of test-set leakage for a paper whose central claim is domain generalization. The differences between the chosen settings and nearby alternatives are small, but the claim would be stronger if the hyperparameters were selected on a validation split (e.g., a held-out subset of REAL275 training videos or using Wild6D) or if a sensitivity analysis showed that the reported margins are robust. Please address this issue.
  3. [Tabs. 1-9] All experimental results are single-run point estimates. Several comparisons in Tab. 1 are separated by small margins (e.g., 3D50: 69.2 vs. 67.2 for DPDN; 5°2cm: 36.5 vs. 30.2 for TTA-COPE), and the Wild6D comparison in Tab. 3 also relies on single numbers. Without standard deviations or multiple seeds, it is not possible to assess whether the reported improvements are statistically significant. Please report mean and standard deviation over at least three training runs, or otherwise justify the stability of the results.
minor comments (6)
  1. [Sec. 5.3.1] The text says Diff9D achieves 43.9% and 54.8% on 5°5cm and 10°5cm, but in Tab. 1 the 54.8 value corresponds to 10°2cm (and 70.0 to 10°5cm); the metric names in this sentence appear to be swapped.
  2. [Eq. (7)] The symbol X_θ appears to be a typo for the variance Σ_θ.
  3. [Sec. 3.1.1] The phrase 'sampled from a real-world pose distribution' should read 'sampled from the data distribution', since the method is trained on synthetic poses.
  4. [Sec. 2.3 and Sec. 5.5.5] The claim that Markov-chain sampling 'makes the data distribution more uniform' and thereby reduces the domain gap is not directly measured in the paper; the ablation in Tab. 9 shows an overall accuracy gain but does not isolate this mechanism. Consider tempering the claim or providing a quantitative analysis.
  5. [Sec. 5.3.1] The statement that UDA-COPE 'requires real-world mask labels for learning' appears inconsistent with Tab. 1, which lists UDA-COPE as using real-world data without labels; please clarify.
  6. [Sec. 5.2] The DDIM sampling schedule is described as taking samples every 333 time steps for T=1000 and S=3; since 1000/3 is not an integer, please clarify the exact schedule used.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: the diffusion loss (Eq. 15) trains against externally sampled noise and independent ground-truth poses; the domain-generalization claim is an empirically tested outcome, not a restatement of the loss.

full rationale

The paper's derivation chain is self-contained and non-circular. The forward process (Eqs. 1-5) adds externally sampled Gaussian noise to independent ground-truth 9-DoF pose labels, and the reverse process (Eqs. 6-10) is the standard DDPM/DDIM posterior parameterization, where the network predicts the noise epsilon_t rather than the pose target itself. The training loss in Eq. (15) is a mean-squared-error between the sampled noise and the predicted noise; no evaluation metric or predicted pose is used in defining the supervision, and no fitted parameter is renamed as a prediction. The auxiliary shape and NOCS losses (Eqs. 13-14) supervise auxiliary branches against ground-truth 3D models and NOCS maps from the synthetic training set, again external to the pose evaluation. The claimed domain-generalization benefit is supported by an internal ablation (Table 9) that removes the diffusion components while keeping the same feature extractors and decoder, so the comparison is an empirical outcome rather than a tautology. Self-citations in the paper are related works, comparison baselines, or prior architectural inspirations (e.g., STG6D, DiffPose, MH6D) and are not load-bearing justifications for the central claim; there is no imported uniqueness theorem or ansatz from the authors' own prior work that forces the choice of diffusion. The Euclidean treatment of the rotation matrix in Sec. 3.1 is a modeling assumption that may affect whether the reported n-degree rotation errors are well-defined, but this is a correctness or robustness risk, not circularity: the denoised output is not defined to be the ground-truth pose, nor is the rotation metric used as the training objective. Test-set hyperparameter selection (e.g., S = 3 in Table 8) is a methodological concern for generalization claims, but it does not make any reported quantity equivalent to its input by construction. Therefore, no circular step can be exhibited, and the appropriate score is 0.

Assumptions & free parameters 5 free parameters · 4 assumptions · 0 invented entities

The central method relies on standard deep learning assumptions, including representative training categories and reliable segmentation, plus two paper-specific premises: Euclidean diffusion over rotation matrices, and the uniformization mechanism for domain generalization. No new physical entities or fitted physical constants are introduced; numerical hyperparameters are tuned on the evaluation benchmark.

free parameters (5)
  • Number of DDIM reverse steps S = 3
    Selected from ablation on REAL275 (Tab. 8) as the smallest step count retaining near-optimal accuracy at 17.2 FPS; the headline '3-step near real-time' claim depends on this choice.
  • Transformer depth and skip connections = Tb=7, SC=3
    Chosen from Tab. 7; accuracy saturates near this size, so the reported performance comparisons depend on this architecture choice.
  • Number of attention heads = 16
    Set experimentally in Sec. 5.2, with feature dimension 112 per head; affects the denoiser capacity and results.
  • Diffusion time horizon T and variance schedule = T=1000, schedule values not specified
    Follows DDPM defaults [59]; the forward noise scale and reverse sampling grid are hand-set and affect denoising behavior.
  • Loss weights for shape, NOCS, and pose terms = Implicitly 1.0 each
    Eq. (19) sums Lcd + LS-L1 + Ldiff without weights; no sensitivity analysis is reported, though relative weighting affects what the network optimizes.
assumptions (4)
  • domain assumption Gaussian diffusion in Euclidean R^15 is a valid model for pose vectors that include a rotation matrix (SO(3)).
    Eqs. (1)-(5) add isotropic Gaussian noise to all 15 pose coordinates, and DDIM denoising Eq. (10) operates on the same Euclidean space; no projection or Lie-group treatment is used (Sec. 3.1).
  • domain assumption Training only on CAMERA25 synthetic renders transfers to the real-world object categories in REAL275 and Wild6D.
    The domain-generalization claim is evaluated only within the six CAMERA25 categories (five in Wild6D); the premise that these synthetic shapes and appearances are representative of real intra-class variation is not independently verified.
  • domain assumption Instance segmentation (Mask R-CNN) is available and sufficiently accurate in the target domain.
    The pose pipeline operates on crops and point clouds produced by a pretrained or synthetic-retrained Mask R-CNN; segmentation errors are not modeled (Sec. 3, Fig. 3, Sec. 5.2).
  • ad hoc to paper Markov-chain sampling makes the learned pose distribution more uniform and thereby reduces the synthetic-to-real domain gap.
    This mechanism is asserted in Sec. 2.3 and 5.5.5 with citation [48]; the support is an ablation showing diffusion helps, not a direct measurement of distribution uniformity or a derivation linking uniformity to pose accuracy.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Diff9D: Diffusion-Based Domain-Generalized Category-Level 9-DoF Object Pose Estimation." pith.science (2026). https://pith.science/paper/N5D5H7LF

@misc{pith2026250202525,
  author       = {Pith},
  title        = {Pith review of: Diff9D: Diffusion-Based Domain-Generalized Category-Level 9-DoF Object Pose Estimation},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/N5D5H7LF}},
  note         = {Machine review of arXiv:2502.02525}
}
read the original abstract

Nine-degrees-of-freedom (9-DoF) object pose and size estimation is crucial for enabling augmented reality and robotic manipulation. Category-level methods have received extensive research attention due to their potential for generalization to intra-class unknown objects. However, these methods require manual collection and labeling of large-scale real-world training data. To address this problem, we introduce a diffusion-based paradigm for domain-generalized category-level 9-DoF object pose estimation. Our motivation is to leverage the latent generalization ability of the diffusion model to address the domain generalization challenge in object pose estimation. This entails training the model exclusively on rendered synthetic data to achieve generalization to real-world scenes. We propose an effective diffusion model to redefine 9-DoF object pose estimation from a generative perspective. Our model does not require any 3D shape priors during training or inference. By employing the Denoising Diffusion Implicit Model, we demonstrate that the reverse diffusion process can be executed in as few as 3 steps, achieving near real-time performance. Finally, we design a robotic grasping system comprising both hardware and software components. Through comprehensive experiments on two benchmark datasets and the real-world robotic system, we show that our method achieves state-of-the-art domain generalization performance. Our code will be made public at https://github.com/CNJianLiu/Diff9D.

Figures

Figures reproduced from arXiv: 2502.02525 by the authors.

Figure 1
Figure 1. Comparison of diffusion model-based image generation and object pose estimation. (a): The process of diffusion model, where q and pθ represent the forward (noising) and reverse (denoising) diffusion processes, respectively. (b): Diffusion model-based image generation task, which generates an image based on a prompt. (c): The Overall pipeline of our Diff9D, which redefines the 9-DoF object pose and size estimation ta… view at source ↗
Figure 2
Figure 2. Some visualizations of the reverse diffusion process, representing the diffusion from Gaussian noise poses to objects poses in the observed [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. Workflow of the proposed Diff9D, which includes three main parts (pose diffusion, condition extraction for pose diffusion, and transformer [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗
Figures from the paper (8 more)
Figure 4
Figure 4. Figure 4: Detailed architecture of the shape estimator and shape encoder. [PITH_FULL_IMAGE:figures/full_fig_p005_4.png]
Figure 5
Figure 5. Figure 5: Some visualizations of the estimated shape and NOCS shape. Top, middle, and bottom rows denote the observed RGB images and their [PITH_FULL_IMAGE:figures/full_fig_p006_5.png]
Figure 6
Figure 6. Figure 6: Detailed architecture of the Transformer block. Left: Transformer [PITH_FULL_IMAGE:figures/full_fig_p007_6.png]
Figure 8
Figure 8. Figure 8: Illustration of the robotic grasping strategy. Left and Middle: [PITH_FULL_IMAGE:figures/full_fig_p007_8.png]
Figure 9
Figure 9. Figure 9: Qualitative comparison results on the real-world REAL275 dataset. Both SGPA [ [PITH_FULL_IMAGE:figures/full_fig_p009_9.png]
Figure 10
Figure 10. Figure 10: Qualitative comparison on the real-world Wild6D dataset. SGPA [ [PITH_FULL_IMAGE:figures/full_fig_p011_10.png]
Figure 11
Figure 11. Figure 11: Qualitative comparison results in four real-world robotic grasping scenes: general, occlusion, dark, and clutter. Each test scene contains [PITH_FULL_IMAGE:figures/full_fig_p012_11.png]
Figure 13
Figure 13. Figure 13: Failure cases visualization. Arrows point to focused areas. [PITH_FULL_IMAGE:figures/full_fig_p012_13.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

69 extracted references · 63 canonical work pages

  1. [1]

    Deep learning-based ob- ject pose estimation: A comprehensive survey,

    J. Liu, W. Sun, H. Yang, Z. Zeng, C. Liu, J. Zheng, X. Liu, H. Rahmani, N. Sebe, and A. Mian, “Deep learning-based ob- ject pose estimation: A comprehensive survey,” in arXiv preprint arXiv:2405.07801, 2024

    arXiv 2024

  2. [2]

    Occlusion- aware self-supervised monocular 6d object pose estimation,

    G. Wang, F. Manhardt, X. Liu, X. Ji, and F. Tombari, “Occlusion- aware self-supervised monocular 6d object pose estimation,” in IEEE. Trans. Pattern. Anal. Mach. Intell., vol. 46, no. 3, pp. 1788–1803, 2024

    work page 2024

  3. [3]

    Match normalization: Learning-based point cloud registration for 6d object pose estima- tion in the real world,

    Z. Dang, L. Wang, Y. Guo, and M. Salzman, “Match normalization: Learning-based point cloud registration for 6d object pose estima- tion in the real world,” in IEEE. Trans. Pattern. Anal. Mach. Intell. , early access, pp. 1–15, 2024, doi: 10.1109/TPAMI.2024.3355198

    arXiv 2024

  4. [4]

    Rnnpose: 6-dof ob- ject pose estimation via recurrent correspondence field estimation and pose optimization,

    Y. Xu, K. -Y. Lin, G. Zhang, X. Wang, and H. Li, “Rnnpose: 6-dof ob- ject pose estimation via recurrent correspondence field estimation and pose optimization,” in IEEE. Trans. Pattern. Anal. Mach. Intell. , early access, pp. 1–15, 2024, doi: 10.1109/TPAMI.2024.3360181

    work page arXiv 2024

  5. [5]

    Dpodv2: Dense correspondence-based 6 dof pose estimation,

    I. Shugurov, S. Zakharov, and S. Ilic, “Dpodv2: Dense correspondence-based 6 dof pose estimation,” in IEEE. Trans. Pat- tern. Anal. Mach. Intell., vol. 44, no. 11, pp. 7417–7435, 2022

    work page 2022

  6. [6]

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

    S. Peng, X. Zhou, Y. Liu, H. Lin, Q. Huang, and H. Bao, “Pvnet: Pixel-wise voting network for 6dof object pose estimation,” inIEEE. Trans. Pattern. Anal. Mach. Intell., vol. 44, pp. 3212–3223, 2022

    work page 2022

  7. [7]

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

    H. Chen, W. Tian, P . Wang, F. Wang, L. Xiong, and H. Li, “Epro- pnp: Generalized end-to-end probabilistic perspective-n-points for monocular object pose estimation,” in IEEE. Trans. Pattern. Anal. Mach. Intell., early access, pp. 1–12, 2024

    work page 2024

  8. [8]

    Pvn3d: A deep point-wise 3d keypoints voting network for 6dof pose estimation,

    Y. He, W. Sun, H. Huang, J. Liu, H. Fan, and J. Sun, “Pvn3d: A deep point-wise 3d keypoints voting network for 6dof pose estimation,” in Proc. IEEE Conf. Comput. Vis. Pattern Recognit. (CVPR) , 2020, pp. 11632–11641

    work page 2020

Show all 69 references
  1. [9]

    Vote from the center: 6 dof pose estimation in rgb-d images by radial keypoint voting,

    Y. Wu, M. Zand, A. Etemad, and M. Greenspan, “Vote from the center: 6 dof pose estimation in rgb-d images by radial keypoint voting,” in Proc. Eur. Conf. Comput. Vis. (ECCV), 2022, pp. 335–352

    2022

  2. [10]

    Learning-based point cloud registration for 6d object pose estimation in the real world,

    Z. Dang, L. Wang, Y. Guo, and M. Salzmann, “Learning-based point cloud registration for 6d object pose estimation in the real world,” in Proc. Eur. Conf. Comput. Vis. (ECCV), 2022, pp. 19–37

    2022

  3. [11]

    Center- based decoupled point-cloud registration for 6d object pose estima- tion,

    H. Jiang, Z. Dang, S. Gu, J. Xie, M. Salzmann, and J. Yang, “Center- based decoupled point-cloud registration for 6d object pose estima- tion,” in Proc. IEEE Int. Conf. Comput. Vis. (ICCV) , 2023, pp. 3427– 3437

    2023

  4. [12]

    Implicit 3d orientation learning for 6d object detection from rgb images,

    M. Sundermeyer, Z-C. Marton, M. Durner, M. Brucker, and R. Triebel, “Implicit 3d orientation learning for 6d object detection from rgb images,” in Proc. Eur. Conf. Comput. Vis. (ECCV) , 2018, pp. 699–715

    2018

  5. [13]

    Pose-guided auto-encoder and feature-based refinement for 6-dof object pose regression,

    Z. Li, and X. Ji, “Pose-guided auto-encoder and feature-based refinement for 6-dof object pose regression,” in Proc. IEEE Int. Conf. Robot. Autom. (ICRA), 2020, pp. 8397–8403

    2020

  6. [14]

    Densefusion: 6d object pose estimation by iterative dense fusion,

    C. Wang, D. Xu, Y. Zhu, R. Mart ´ın-Mart´ın, C. Lu, L. Fei-Fei, and S. Savarese, “Densefusion: 6d object pose estimation by iterative dense fusion,” in Proc. IEEE Conf. Comput. Vis. Pattern Recognit. (CVPR), 2019, pp. 3343–3352

    2019

  7. [15]

    Ffb6d: A full flow bidirectional fusion network for 6d pose estimation,

    Y. He, H. Huang, H. Fan, Q. Chen, and J. Sun, “Ffb6d: A full flow bidirectional fusion network for 6d pose estimation,” in Proc. IEEE Conf. Comput. Vis. Pattern Recognit. (CVPR), 2021, pp. 3003–3013

    2021

  8. [16]

    Gdr-net: Geometry- guided direct regression network for monocular 6d object pose estimation ,

    G. Wang, F. Manhardt, F. Tombari, and X. Ji, “Gdr-net: Geometry- guided direct regression network for monocular 6d object pose estimation ,” in Proc. IEEE Conf. Comput. Vis. Pattern Recognit. (CVPR), 2021, pp. 16611–16621

    2021

  9. [17]

    Hff6d: Hier- archical feature fusion network for robust 6d object pose tracking,

    J. Liu, W. Sun, C. Liu, X. Zhang, S. Fan, and W. Wu, “Hff6d: Hier- archical feature fusion network for robust 6d object pose tracking,” IEEE Trans. Circuits Syst. Video Technol. , vol. 32, no. 11, pp. 7719– 7731, 2022

    2022

  10. [18]

    Mfpn-6d : Real- time one-stage pose estimation of objects on rgb images,

    P . Liu, Q. Zhang, J. Zhang, F. Wang, and J. Cheng, “Mfpn-6d : Real- time one-stage pose estimation of objects on rgb images,” in Proc. IEEE Int. Conf. Robot. Autom. (ICRA) , 2021, pp. 12939–12945

    2021

  11. [19]

    Normalized object coordinate space for category-level 6d object pose and size estimation,

    H. Wang, S. Sridhar, J. Huang, J. Valentin, S. Song, and L. J. Guibas, “Normalized object coordinate space for category-level 6d object pose and size estimation,” in Proc. IEEE Conf. Comput. Vis. Pattern Recognit. (CVPR), 2019, pp. 2642–2651

    2019

  12. [20]

    Shape prior deformation for categorical 6d object pose and size estimation,

    M. Tian, M. H. Ang, and G. H. Lee, “Shape prior deformation for categorical 6d object pose and size estimation,” in Proc. Eur. Conf. Comput. Vis. (ECCV), 2020, pp. 530–546

    2020

  13. [21]

    Category-level 6d object pose estimation via cascaded relation and recurrent reconstruction net- works,

    J. Wang, K. Chen, and Q. Dou, “Category-level 6d object pose estimation via cascaded relation and recurrent reconstruction net- works,” in Proc. IEEE/RSJ Int. Conf. Intell. Robot. Syst. (IROS) , 2021, pp. 4807–4814

    2021

  14. [22]

    Sgpa: Structure-guided prior adaptation for category-level 6d object pose estimation,

    K. Chen, and Q. Dou, “Sgpa: Structure-guided prior adaptation for category-level 6d object pose estimation,” inProc. IEEE Int. Conf. Comput. Vis. (ICCV), 2021, pp. 2773–2782

    2021

  15. [23]

    6d-vit: Category-level 6d object pose estimation via transformer-based instance representa- tion learning,

    L. Zou, Z. Huang, N. Gu, and G. Wang, “6d-vit: Category-level 6d object pose estimation via transformer-based instance representa- tion learning,” in IEEE Trans. Image Process., vol. 31, pp. 6907–6921, 2022

    2022

  16. [24]

    Robotic contin- uous grasping system by shape transformer-guided multiobject category-level 6-d pose estimation,

    J. Liu, W. Sun, C. Liu, X. Zhang, and Q. Fu, “Robotic contin- uous grasping system by shape transformer-guided multiobject category-level 6-d pose estimation,” in IEEE Trans. Ind. Informat. , vol. 32, no. 10, pp. 6728–6740, 2023

    2023

  17. [25]

    Rbp- pose: Residual bounding box projection for category-level pose estimation,

    R. Zhang, Y. Di, Z. Lou, F. Manhardt, F. Tombari, and X. Ji, “Rbp- pose: Residual bounding box projection for category-level pose estimation,” in Proc. Eur. Conf. Comput. Vis. (ECCV), 2022, pp. 655– 672

    2022

  18. [26]

    Catre: Iterative point clouds alignment for category-level object pose refinement,

    X. Liu, G. Wang, Y. Li, and X. Ji, “Catre: Iterative point clouds alignment for category-level object pose refinement,” in Proc. Eur. Conf. Comput. Vis. (ECCV), 2022, pp. 499–516

    2022

  19. [27]

    Georef: Geometric alignment across shape variation for category-level object pose refinement,

    L. Zheng, T. H. E. Tse, C. Wang, Y. Sun, H. Chen, A. Leonardis, and W. Zhang, “Georef: Geometric alignment across shape variation for category-level object pose refinement,” in Proc. IEEE Conf. Comput. Vis. Pattern Recognit. (CVPR), 2024, pp. 10693–10703

    2024

  20. [28]

    Dualposenet: Category- level 6d object pose and size estimation using dual pose network with refined learning of pose consistency,

    J. Lin, Z. Wei, Z. Li, S. Xu, K. Jia, and Y. Li, “Dualposenet: Category- level 6d object pose and size estimation using dual pose network with refined learning of pose consistency,” in Proc. IEEE Int. Conf. Comput. Vis. (ICCV), 2021, pp. 3560–3569

    2021

  21. [29]

    Fs-net: Fast shape-based network for category-level 6d object pose estimation with decoupled rotation mechanism,

    W. Chen, X. Jia, H. J. Chang, J. Duan, L. Shen, and A. Leonardis, “Fs-net: Fast shape-based network for category-level 6d object pose estimation with decoupled rotation mechanism,” inProc. IEEE Conf. Comput. Vis. Pattern Recognit. (CVPR), 2021, pp. 1581–1590. IEEE TRANSACTIONS...

    2021

  22. [30]

    Gpv-pose: Category-level object pose estimation via geometry-guided point-wise voting,

    Y. Di, R. Zhang, Z. Lou, F. Manhardt, X. Ji, N. Navab, and F. Tombari, “Gpv-pose: Category-level object pose estimation via geometry-guided point-wise voting,” in Proc. IEEE Conf. Comput. Vis. Pattern Recognit. (CVPR), 2022, pp. 6781–6791

    2022

  23. [31]

    Hs-pose: Hybrid scope feature extrac- tion for category-level object pose estimation,

    L. Zheng, C. Wang, Y. Sun, E. Dasgupta, H. Chen, A. Leonardis, W. Zhang, and H. J. Chang, “Hs-pose: Hybrid scope feature extrac- tion for category-level object pose estimation,” in Proc. IEEE Conf. Comput. Vis. Pattern Recognit. (CVPR), 2023, pp. 17163–17173

    2023

  24. [32]

    Ist-net: Prior-free category-level pose estimation with implicit space transformation,

    J. Liu, Y. Chen, X. Ye, and X. Qi, “Ist-net: Prior-free category-level pose estimation with implicit space transformation,” in Proc. IEEE Int. Conf. Comput. Vis. (ICCV), 2023, pp. 13978–13988

    2023

  25. [33]

    Vi-net: Boosting category- level 6d object pose estimation via learning decoupled rotations on the spherical representations,

    J. Lin, Z. Wei, Y. Zhang, and K. Jia, “Vi-net: Boosting category- level 6d object pose estimation via learning decoupled rotations on the spherical representations,” in Proc. IEEE Int. Conf. Comput. Vis. (ICCV), 2023, pp. 14001–14011

    2023

  26. [34]

    Cppf: Towards robust category-level 9d pose estimation in the wild,

    Y. You, R. Shi, W. Wang, and C. Lu, “Cppf: Towards robust category-level 9d pose estimation in the wild,” in Proc. IEEE Conf. Comput. Vis. Pattern Recognit. (CVPR), 2022, pp. 6866–6875

    2022

  27. [35]

    Sar-net: Shape alignment and recovery network for category-level 6d object pose and size estimation,

    H. Lin, Z. Liu, C. Cheang, Y. Fu, G. Guo, and X. Xue, “Sar-net: Shape alignment and recovery network for category-level 6d object pose and size estimation,” in Proc. IEEE Conf. Comput. Vis. Pattern Recognit. (CVPR), 2022, pp. 6707–6717

    2022

  28. [36]

    Self-supervised category- level 6d object pose estimation with deep implicit shape represen- tation,

    W. Peng, J. Yan, H. Wen, and Y. Sun, “Self-supervised category- level 6d object pose estimation with deep implicit shape represen- tation,” in Proc. AAAI Conf. Artif. Intell. (AAAI) , vol. 36, no. 2, 2022, pp. 2082–2090

    2022

  29. [37]

    Deepsdf: Learning continuous signed distance functions for shape representation,

    J. J. Park, P . Florence, J. Straub, R. Newcombe, and S. Lovegrove, “Deepsdf: Learning continuous signed distance functions for shape representation,” in Proc. IEEE Conf. Comput. Vis. Pattern Recognit. (CVPR), 2019, pp. 165–174

    2019

  30. [38]

    Udacope: Unsupervised domain adaptation for category- level object pose estimation,

    T. Lee, B. -U. Lee, I. Shin, J. Choe, U. Shin, I. S. Kweon, and K. -J. Yoon, “Udacope: Unsupervised domain adaptation for category- level object pose estimation,” inProc. IEEE Conf. Comput. Vis. Pattern Recognit. (CVPR), 2022, pp. 14891–14900

    2022

  31. [39]

    Category-level 6d object pose estimation in the wild: A semi-supervised learning approach and a new dataset,

    Y. Fu, and X. Wang, “Category-level 6d object pose estimation in the wild: A semi-supervised learning approach and a new dataset,” in Proc. Adv. Neural Inf. Process. Syst. (NeurIPS), 2022, vol. 35

    2022

  32. [40]

    Category-level 6d object pose and size estimation using self-supervised deep prior deformation networks,

    J. Lin, Z. Wei, C. Ding, and K. Jia, “Category-level 6d object pose and size estimation using self-supervised deep prior deformation networks,” in Proc. Eur. Conf. Comput. Vis. (ECCV), 2022, pp. 19–34

    2022

  33. [41]

    Tta-cope: Test-time adap- tation for category-level object pose estimation,

    T. Lee, J. Tremblay, V . Blukis, B. Wen, B. -U. Lee, I. Shin, S. Birchfield, I. S. Kweon, and K. -J. Yoon, “Tta-cope: Test-time adap- tation for category-level object pose estimation,” in Proc. IEEE Conf. Comput. Vis. Pattern Recognit. (CVPR), 2023, pp. 21285–21295

    2023

  34. [42]

    Towards unsupervised domain generalization,

    X. Zhang, L. Zhou, R. Xu, P . Cui, Z. Shen, and H. Liu, “Towards unsupervised domain generalization,” in Proc. IEEE Conf. Comput. Vis. Pattern Recognit. (CVPR), 2022, pp. 4910–4920

    2022

  35. [43]

    Denoising diffusion implicit models,

    J. Song, C. Meng, and S. Ermon, “Denoising diffusion implicit models,” in Proc. Int. Conf. Learn. Representations (ICLR), 2021

    2021

  36. [44]

    Ssd- 6d: Making rgb-based 3d detection and 6d pose estimation great again,

    W. Kehl, F. Manhardt, F. Tombari, S. Ilic, and N. Navab, “Ssd- 6d: Making rgb-based 3d detection and 6d pose estimation great again,” in Proc. IEEE Int. Conf. Comput. Vis., 2017, pp. 1530–1538

    2017

  37. [45]

    Se (3) diffusion model-based point cloud registration for robust 6d object pose estimation,

    H. Jiang, M. Salzmann, Z. Dang, J. Xie, and J. Yang, “Se (3) diffusion model-based point cloud registration for robust 6d object pose estimation,” in Proc. Adv. Neural Inf. Process. Syst. (NeurIPS) , 2023, vol. 36

    2023

  38. [46]

    6d-diff: A keypoint diffusion framework for 6d object pose estimation,

    L. Xu, H. Qu, Y. Cai, and J. Liu, “6d-diff: A keypoint diffusion framework for 6d object pose estimation,” in Proc. IEEE Conf. Comput. Vis. Pattern Recognit. (CVPR), 2024, pp. 9676–9686

    2024

  39. [47]

    Diffusionnocs: Managing symmetry and uncertainty in sim2real multi-modal category-level pose esti- mation,

    T. Ikeda, S. Zakharov, T. Ko, M. Z. Irshad, R. Lee, K. Liu, R. Ambrus, and K. Nishiwaki, “Diffusionnocs: Managing symmetry and uncertainty in sim2real multi-modal category-level pose esti- mation,” in Proc. IEEE Int. Conf. Robot. Autom. (ICRA) , 2024

    2024

  40. [48]

    Ed-sam: An efficient diffusion sampling approach to domain generaliza- tion in vision-language foundation models,

    T. D. Truong, X. Li, B. Raj, J. Cothren, and K. Luu, “Ed-sam: An efficient diffusion sampling approach to domain generaliza- tion in vision-language foundation models,” in arXiv preprint arXiv:2406.01432, 2024

    2024 arXiv

  41. [49]

    Deep residual learning for image recognition,

    K. He, X. Zhang, S. Ren, and J. Sun, “Deep residual learning for image recognition,” in Proc. IEEE Conf. Comput. Vis. Pattern Recognit. (CVPR), 2016, pp. 770–778

    2016

  42. [50]

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

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

    2017

  43. [51]

    A practical robotic grasping method by using 6-d pose estimation with protective correction,

    H. Zhang, Z. Liang, C. Li, H. Zhong, L. Liu, C. Zhao, Y. Wang, and Q. M. J. Wu, “A practical robotic grasping method by using 6-d pose estimation with protective correction,” in IEEE Trans. Ind. Electron., vol. 69, no. 4, pp. 3876-3886, 2022

    2022

  44. [52]

    Genpose: Generative category- level object pose estimation via diffusion models,

    J. Zhang, M. Wu, and H. Dong, “Genpose: Generative category- level object pose estimation via diffusion models,” in Proc. Adv. Neural Inf. Process. Syst. (NeurIPS), 2023, vol. 36, pp. 1–14

    2023

  45. [53]

    Fine segmenta- tion and difference-aware shape adjustment for category-level 6dof object pose estimation,

    C. Liu, W. Sun, J. Liu, X. Zhang, S. Fan, and Q. Fu, “Fine segmenta- tion and difference-aware shape adjustment for category-level 6dof object pose estimation,” in Appl. Intell. , vol. 53, pp. 23711–23728, 2023

    2023

  46. [54]

    Wolnitza, O

    M. Wolnitza, O. Kaya, T. Kulvicius, F. Worgotter, and B. Dellen, ”6D pose estimation and 3d object reconstruction from 2d shape for robotic grasping of objects,” in Proc. IEEE Int. Conf. Robot. Comput. (IRC), 2022, pp. 67–71

    2022

  47. [55]

    Bdr6d: Bidirectional deep residual fusion network for 6d pose estimation,

    P . Liu, Q. Zhang, and J. Cheng, “Bdr6d: Bidirectional deep residual fusion network for 6d pose estimation,” in IEEE Trans. Autom. Sci. Eng., vol. 21, no. 2, pp. 1793–1804, 2024

    2024

  48. [56]

    J. Liu, W. Sun, H. Yang, C. Liu, X. Zhang, and A. Mian, ”Domain- generalized robotic picking via contrastive learning-based 6-d pose estimation,” in IEEE Trans. Ind. Informat. , vol. 20, no. 6, pp. 8650– 8661, 2024

    2024

  49. [57]

    Robotic grasp detection based on category-level object pose estimation with self-supervised learn- ing,

    S. Yu, D. -H. Zhai, and Y. Xia, “Robotic grasp detection based on category-level object pose estimation with self-supervised learn- ing,” in IEEE Trans. Mechatronics, vol. 29, no. 1, pp. 625–635, 2024

    2024

  50. [58]

    Sim-to-real 6d object pose estimation via iterative self-training for robotic bin picking,

    K. Chen, R. Cao, S. James, Y. Li, Y. -H. Liu, P . Abbeel, and Q. Dou, “Sim-to-real 6d object pose estimation via iterative self-training for robotic bin picking,” in Proc. Eur. Conf. Comput. Vis. (ECCV) , 2022, pp. 533–550

    2022

  51. [59]

    Denoising diffusion probabilistic models,

    J. Ho, A. Jain, and P . Abbeel, “Denoising diffusion probabilistic models,” in Proc. Adv. Neural Inf. Process. Syst. (NeurIPS) , 2020, vol. 33, pp. 6840–6851

    2020

  52. [60]

    Mask r-cnn,

    K. He, G. Gkioxari, P . Dollar, and R Girshick, “Mask r-cnn,” in Proc. IEEE Int. Conf. Comput. Vis. (ICCV), 2017, pp. 2961–2969

    2017

  53. [61]

    Diffpose: Toward more reliable 3d pose estimation,

    J. Gong, L. G. Foo, Z. Fan, Q. Ke, H. Rahmani, and J. Liu, “Diffpose: Toward more reliable 3d pose estimation,” in Proc. IEEE Conf. Comput. Vis. Pattern Recognit. (CVPR), 2023, pp. 13041–13051

    2023

  54. [62]

    Shapenet: An information-rich 3d model repository,

    A. X. Chang, T. Funkhouser, L. Guibas, P . Hanrahan, Q. Huang, Z. Li, S. Savarese, M. Savva, S. Song, H. Su, J. Xiao, L. Yi, and F. Yu, “Shapenet: An information-rich 3d model repository,” in arXiv preprint arXiv:1512.03012, 2015

    2015 arXiv

  55. [63]

    Secondpose: Se (3)-consistent dual-stream feature fusion for category-level pose estimation,

    Y. Chen, D. Yan, G. Zhai, F. Manhardt, C. Zhang, R. Zhang, F. Tombari, N. Navab, and B. Busam, “Secondpose: Se (3)-consistent dual-stream feature fusion for category-level pose estimation,” in Proc. IEEE Conf. Comput. Vis. Pattern Recognit. (CVPR) , 2024, pp. 9959–9969

    2024

  56. [64]

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

    A. Dosovitskiy, L. Beyer, A. Kolesnikov, et al., “An image is worth 16x16 words: Transformers for image recognition at scale,” in Proc. Int. Conf. Learn. Representations (ICLR), 2021

    2021

  57. [65]

    P . Deng, C. Ge, X. Qiao, H. Wei, and Y. Sun, ”Attention-aware dual-stream network for multimodal face anti-spoofing,” in IEEE Trans. Inf. Forensics Secur., vol. 18, pp. 4258–4271, 2023

    2023

  58. [66]

    Cyclical learning rates for training neural networks,

    L. N. Smith, “Cyclical learning rates for training neural networks,” in Proc. IEEE Winter Conf. Appl. Comput. Vis. (WACV), 2017, pp. 464– 472

    2017

  59. [67]

    Mh6d: Multi-hypothesis consistency learning for category-level 6-d object pose estimation,

    J. Liu, W. Sun, C. Liu, H. Yang, X. Zhang, and A. Mian, “Mh6d: Multi-hypothesis consistency learning for category-level 6-d object pose estimation,” in IEEE Trans. Neural Netw. Learn. Syst. , early access, pp. 1–14, 2024, doi: 10.1109/TNNLS.2024.3360712

    2024

  60. [68]

    High-resolution image synthesis with latent diffusion models,

    R. Rombach, A. Blattmann, D. Lorenz, P . Esser, and B. Ommer, “High-resolution image synthesis with latent diffusion models,” in Proc. IEEE Conf. Comput. Vis. Pattern Recognit. (CVPR) , 2022, pp. 10684–10695

    2022

  61. [69]

    Dpm-solver++: Fast solver for guided sampling of diffusion probabilistic models,

    C. Lu, Y. Zhou, F. Bao, J. Chen, C. Li, and J. Zhu, “Dpm-solver++: Fast solver for guided sampling of diffusion probabilistic models,” in arXiv preprint arXiv:2211.01095, 2022. Jian Liu is currently pursuing his Ph.D. degree at the National Engineering Research Center of Robot...

    2022 arXiv

Pith tools

Reviewed August 9, 2026 · model on record in the stance chip above.