Pith. sign in

REVIEW 3 major objections 7 minor 55 references

GTR: Gaussian Splatting Tracking and Reconstruction of Unknown Objects Based on Appearance and Geometric Complexity

T0 review · 3 major / 7 minor · reviewed 2026-08-15 · deepseek-v4-flash

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

Pith's one-line read This paper claims that an adaptive pipeline combining 3D Gaussian splatting, complexity-gated keypoint tracking, keyframe selection, and pose graph optimization achieves consistent 6-DoF tracking and complete 3D reconstruction of unknown…

desk verdict GTR is a competent adaptive tracking-plus-reconstruction system with a useful new benchmark; the synthetic evidence is strong, but the real-data comparison for axis-symmetric objects rests on unvalidated reference poses. read the letter →

arxiv 2505.11905 v1 pith:VENISOM6 submitted 2025-05-17 cs.CV cs.RO

Takuya Ikeda , Sergey Zakharov , Muhammad Zubair Irshad , Istvan Balazs Opra , Shun Iwase , Dian Chen , Mark Tjersland , Robert Lee
show 3 more authors
Alexandre Dilly Rares Ambrus Koichi Nishiwaki
This is my paper · ORCID
classification cs.CVcs.RO
keywords 6-DoFobjecttracking3DGaussianSplattingRGB-Dreconstructionposegraphoptimizationaxis-symmetricobjectskeyframeselectionappearancecomplexitygeometric
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 sets out to show that a single-sensor RGB-D system can track and reconstruct unknown household objects in full 360 degrees, even when those objects are axis-symmetric, low-texture, or geometrically intricate. It argues that the key is to make the pipeline adapt to the object's appearance and geometry instead of using one fixed tracking recipe. To that end it introduces adaptive tracker selection, a complexity-gated 3D Gaussian splatting pose refiner, and a keyframe scheme based on visibility and rotation, and it evaluates these on a new six-object benchmark with synthetic and real sequences. The central claim is that this combination yields the lowest average reconstruction error and competitive pose accuracy, with particularly large gains over BundleSDF on symmetric bottles and cans.

What carries the argument

The central mechanism is the complexity-gated adaptive pipeline that decides, per frame, which tracker and which pose refiner to use. Appearance complexity is quantified by SIFT feature density, which routes tracking to either TAP or LoFTR. Geometric complexity is computed as the mean of per-point local curvature $\kappa_i = \lambda_1/(\lambda_1+\lambda_2+\lambda_3)$ from the eigenvalues of the local covariance matrix, and this value gates whether the 3D Gaussian Splatting render-and-compare refiner (using color and depth losses) is applied. A keyframe pool is updated using rotational geodesic distance and a keypoint visibility rate, and pose graph optimization closes loops when visibility and geodesic conditions are met. These components together transform coarse correspondences from TEASER++ and ICP into a globally consistent trajectory that supports TSDF mesh extraction.

What would settle it

Re-annotate the real GTR3D can and oil-bottle sequences using a motion-capture turntable or fiducial markers that provide independent 6-DoF ground truth, then recompute the rotation and translation errors for GTR, BundleSDF, and Gaussian-SLAM; if the new ground-truth poses differ significantly from the retained FoundationPose results, the reported real-data rankings for those two rows in Table 3 may change.

Watch

Extended reading notes

Core claim

The authors claim that their method, GTR, outperforms state-of-the-art joint tracking-and-reconstruction baselines on the GTR3D benchmark. On synthetic data, it achieves the lowest Chamfer distance for every object, with mean CD 1.83 versus 4.03 for BundleSDF and 48.83 for Gaussian-SLAM, and it dramatically reduces rotation error on axis-symmetric objects: 33.43° and 1.94° for the can and oil bottle versus 90.39° and 92.38° for BundleSDF. On real sequences, it again posts the best average reconstruction error (CD 12.01) while keeping translation and rotation errors in a similar range as Gaussian-SLAM. The method's design is adaptive: it measures appearance complexity by SIFT feature density to choose between a learned keypoint tracker and a low-texture feature matcher, measures geometric complexity by mean local curvature to decide whether to run an expensive 3DGS render-and-compare refinement, and uses visibility plus geodesic distance for keyframe selection and loop-closure-driven pose graph optimization.

Load-bearing premise

The real-world ground-truth poses for the two axis-symmetric objects come from a model-based tracker that the paper itself does not independently verify for yaw motion, and the alternative tracker was discarded for being unstable on exactly that motion, so the measured errors on those objects rest on an unverified estimate of the true pose.

Editorial extensions

If this is right

  • A complete object mesh can be recovered from a single RGB-D camera without any CAD model, making object scanning for robotics and AR more accessible.
  • Axis-symmetric objects, previously a known failure case for neural joint tracking, can be tracked through full yaw rotations when 3DGS refinement is applied selectively.
  • The GTR3D benchmark, with its synthetic and real full-view sequences for six objects spanning symmetry, texture, and geometry challenges, provides a standardized protocol for comparing future track-and-reconstruct methods.
  • The adaptive gating means computation is spent only where it improves accuracy, so geometrically complex objects can be processed much faster than a uniformly expensive pipeline.
  • Keyframe selection with pose graph optimization reduces drift over long sequences, which is essential for full 360-degree reconstruction from casually captured video.

Reading between the lines

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

  • The real-sequence comparison for the two axis-symmetric objects may be less reliable than the synthetic one, because the retained ground-truth poses come from FoundationPose, a tracker that the paper does not independently validate for yaw motion; re-annotating with motion-capture or fiducial markers could change the reported rotation errors and rankings.
  • The pose graph optimization step is a computational bottleneck (on the order of hours for some sequences), so a real-time or interactive version of the method would need an analytic Jacobian implementation rather than the generic PyPose optimizer.
  • The complexity-gating principle is not tied to 3D Gaussian splatting; the same idea of measuring SIFT density and local curvature to choose trackers and refinement modules could improve other joint tracking and reconstruction systems that use neural fields.
  • Because the method assumes a static camera and moving object, it does not yet address the complementary setting of a moving camera in a static scene; extending the pose graph to include camera motion would connect it to object-level SLAM.
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 / 7 minor

Summary. This paper presents GTR, a monocular-RGBD object-centric 6-DoF tracking and reconstruction system built around 3D Gaussian Splatting. The pipeline first measures appearance complexity from SIFT feature density and switches between TAP and LoFTR-based keypoint tracking; a coarse relative pose is estimated with TEASER++; frames are selected into a keyframe pool using rotational geodesic distance and keypoint visibility rate; poses are refined by ICP and, for frames with low geometric complexity, by a 3DGS render-and-compare step; pose graph optimization is applied on loop closure. A mesh is extracted by fitting a full 3DGS to the keyframes and running TSDF fusion. The authors also introduce the GTR3D benchmark with six objects (juice box, dinosaur toy, spray, drill, energy drink can, oil bottle) chosen to cover axis symmetry, geometric complexity, and texture complexity, each with one synthetic and one real RGB-D sequence. On GTR3D, GTR reports the lowest mean Chamfer distance in both settings (1.83 vs. 4.03 for BundleSDF synthetic; 12.01 vs. 16.02 real) and competitive pose errors. The claimed contributions are an adaptive complexity-aware tracking pipeline and a benchmark targeting axis-symmetric and otherwise challenging objects.

Significance. If the central results hold, the strongest contribution is the adaptive system design: gating dense keypoint tracking (LoFTR vs. TAP) on SIFT density, and gating 3DGS render-and-compare refinement on mean local curvature, is a sensible and clearly described way to allocate computation. The ablation study (Table 4) provides concrete support for the design decisions (3DGS refinement reduces the oil bottle rotation error from 33.59 to 1.94 degrees, and skipping it on geometry-complex objects keeps the dino error low). The GTR3D benchmark addresses a genuine gap in public evaluation: full-view sequences of axis-symmetric objects annotated for tracking and reconstruction, with per-object complexity characterization in Table 1. The paper is also candid about limitations in Appendix F (PGO runtime of roughly two hours per sequence, and unhandled specular/transparent materials). The limitations are in the evidence rather than the algorithm: a single sequence per object per setting, unvalidated reference poses for the real can and oil bottle sequences, and headline claims of 'consistent performance' that are stronger than the tables support.

major comments (3)
  1. [Supplementary B; Table 3] The evidence for the paper's signature claim, robust real-world tracking of axis-symmetric objects, rests on reference poses whose validity is not established. Supplement B reports that ICG+ yaw tracking was 'not stable' for the oil bottle and energy drink can, and that 'we retained the results from FoundationPose' for exactly these two objects; no independent validation of FoundationPose's yaw accuracy on axis-symmetric objects is provided. The large real-data margins in Table 3 (can: GTR 11.37 vs. BundleSDF 67.21; oil: 4.76 vs. 67.09) therefore measure agreement with the FoundationPose reference rather than with true poses. This concern is reinforced by the observation that GTR's can rotation error is 11.37 degrees against the FoundationPose-derived reference in Table 3 but 33.43 degrees against exact ground truth in the synthetic sequence (Table 2). The benchmark's 'high-quality annotations' claim and the real-world robustness claim for symmetry are not supported as reported; please validate the reference (e.g., render-based yaw alignment or a marker-based check), provide a sensitivity analysis, or reframe the real symmetry comparison as an agreement analysis.
  2. [Sec. 4.1; Table 2] The statement in Sec. 4.1 that 'only our method demonstrates consistent performance across all objects' is contradicted by GTR's own results on the synthetic can sequence: Rerr 33.43 degrees and Terr 18.95, the worst translation error in the table and about 4.7 times worse than BundleSDF's Terr of 4.05 on the same object. In the other direction, on the spray and drill objects BundleSDF's rotation error is 4-6 times smaller than GTR's (2.24 vs. 9.35 and 0.58 vs. 3.49). The favorable synthetic means for GTR (Rerr 8.47 vs. 40.59; CD 1.83 vs. 4.03) are largely driven by BundleSDF's near-90-degree failures on the two symmetric objects. This is still a positive result, but the narrative should be reframed as a trade-off favoring axis-symmetric and low-geometry objects, with the can translation error reported prominently.
  3. [Sec. 4; Tables 2-3; Table 4] The empirical support for 'robust tracking' and 'stable performance across all object classes' (Secs. 1 and 4.2) is statistically thin: one sequence per object per setting, no repeated trials, no error bars, and no sensitivity analysis over the numerous thresholds in Sec. 3 (SIFT feature density, geodesic distance thresholds, TAP visibility rates of 90% and 50%, the 20% low-appearance PGO skip fraction, the loss weights in Eq. (5), and the curvature radius and neighbor count). Moreover, on the real tracking metrics in Table 3, GTR is not the best method on the mean pose errors (Gaussian-SLAM: Rerr 5.61 vs. 7.70 and Terr 6.15 vs. 6.37), which the text acknowledges only partially. Reporting per-sequence variance or additional trials, and explicitly positioning GTR as a reconstruction-focused method with competitive-if-not-leading pose accuracy, would make the claims commensurate with the evidence.
minor comments (7)
  1. [Sec. 4.1] The sentence 'We show qualitative results in Table 2' should read 'quantitative results'; the qualitative results appear in Fig. 8.
  2. [Table 4] The table contains at least one copy-paste error: the 'No PGO' Time row (6.04, 1.51, 19.15, 3.12, 58.83, 33.59, 20.37) is identical to the 'No 3DGS' Rerr row, which cannot be correct for reported runtimes; additionally several values are concatenated without separators (e.g., '19.153.1258.83' and '1.948.47'), so the table should be regenerated and proofread.
  3. [Sec. 3.3, Eq. (6)] The phrase 'π−1 j represents the inverse projection mapping from frame Fi' is misstated (it unprojects points from Fi), and the symbol E_{i,j} overloads the pose-graph edge set E defined in the same paragraph.
  4. [Sec. 4.2] The sentence 'while BundleTrack performs well in translation estimation' refers to a baseline (BundleTrack [43]) that does not appear in Tables 2-3; BundleSDF is presumably meant.
  5. [Sec. 3] A complete hyperparameter table is missing: the SIFT feature-density threshold, the geodesic distance threshold, the TAP visibility thresholds (90% and 50%), the 20% low-appearance-complexity PGO skip fraction, the weights λ_color, λ_depth, λ_reg in Eq. (5), and the curvature radius and maximum neighbor count are never given together, which limits reproducibility.
  6. [Sec. 4; Eq. (5) evaluation] The reference geometry used for Chamfer distance is not specified: for the synthetic portion it is presumably the simulation mesh and for the real portion the artist-created CAD model mentioned in Supplement B, but this should be stated explicitly in the main text, including how the two are scaled to a common metric space.
  7. [Throughout] There are several small typos and formatting issues: 'new two keyfames' (Sec. 3.3), 'Bootstap' in reference [4], 'render&comapre' in Supp. F, and the 'BundleTrack' reference in Sec. 4.2 noted above.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the tracking and reconstruction pipeline is assembled from independent components (TAP/LoFTR, TEASER++, ICP, 3DGS render-and-compare, PGO, TSDF fusion) and evaluated with external baselines and independent tracker-generated pose annotations, not with the method's own outputs.

full rationale

The paper's derivation chain is self-contained and does not reduce any claimed prediction to its inputs by construction. The coarse pose comes from keypoint correspondences (TAP or LoFTR) solved by TEASER++, the pose refinement uses ICP and 3DGS render-and-compare with losses defined against observed RGB-D frames (Eqs. 3-5), and global consistency is obtained through pose graph optimization (Eq. 6). None of these steps defines the output pose in terms of the final tracking error or the reconstruction metric. The reconstruction benchmark computes Chamfer Distance after TSDF fusion using recovered poses, and the same procedure is applied uniformly to all baselines, so the comparison is not fitted to make GTR win. Pose annotations for GTR3D Real are produced by external trackers FoundationPose and ICG+ (Supplement B), not by GTR, so the real-data evaluation is not self-referential. Supplement B does contain an important limitation statement: 'the performance of ICG+ in tracking yaw movement (caused by the turntable) for axially symmetric objects was not stable. Consequently, we retained the results from FoundationPose for the axially symmetric objects: the oil bottle and the energy drink can.' This is a measurement-validity caveat for the real-data symmetry comparison, but it is not circularity: it does not make GTR's reported errors equal to any fitted parameter or to the method's own output by construction. The paper's self-citations (e.g., DiffusionNOCS, Shapo, FSD) are background references and are not load-bearing assumptions for the central tracking or reconstruction claims. Therefore the circularity score is 0.

Assumptions & free parameters 6 free parameters · 5 assumptions · 0 invented entities

The method's novelty rests on a set of heuristics and thresholds, none of which are given numerical values or sensitivity analysis. The most consequential assumption is the validity of the ground-truth annotations for the real benchmark, especially for axis-symmetric objects where the annotation method is known to be unstable.

free parameters (6)
  • Appearance complexity threshold (SIFT feature density) = not reported
    Used in Sec. 3.1 to switch between TAP and LoFTR; the threshold value is not given in the paper.
  • Geodesic distance keyframe threshold = not reported
    Controls keyframe insertion and loop closure (Sec. 3.2, 3.3); only '180 degrees minus half' is referenced without the base value.
  • TAP visibility-rate thresholds = 90% and 50%
    PGO is triggered when visibility exceeds 90% (or 50% within 90 degrees geodesic); these are chosen values stated in Sec. 3.3.
  • Low-appearance-complexity PGO skip fraction = 20%
    PGO is skipped if more than 20% of keyframes have low appearance complexity (Sec. 3.3).
  • 3DGS loss weights (lambda_color, lambda_depth, lambda_reg) = not reported
    Weights in Eq. 5 are not specified numerically; lambda in Eq. 4 is given as 0.2.
  • Curvature neighborhood radius and maximum neighbors = not reported
    Parameters for the geometric complexity covariance estimation in Sec. 3.3 are not specified.
assumptions (5)
  • domain assumption The off-the-shelf trackers TAP, LoFTR, SIFT, and SAM2 perform as expected on the benchmark objects.
    The entire coarse pose and keyframe pipeline depends on these pretrained models' zero-shot generalization (Sec. 3.1).
  • domain assumption The learned stereo depth method [29] provides accurate depth maps for the real sequences.
    Real-data depth is generated by a learned stereo model and treated as ground-truth input (Supplement A).
  • domain assumption FoundationPose and ICG+ provide accurate ground-truth poses for the real benchmark.
    Real pose annotations come from these trackers; the paper itself notes ICG+ is unstable for axis-symmetric objects and falls back to FoundationPose (Supplement B).
  • domain assumption 3DGS render-and-compare loss reduces pose error for low-geometric-complexity objects.
    The method assumes Gaussian Splatting optimization is a reliable pose refiner (Sec. 3.3).
  • standard math TEASER++ returns a good coarse pose when correspondences are correct.
    Standard robust registration result used in the coarse pose solver (Sec. 3.1).

how reviews work

0 comments
Cite this review

Pith. "Pith review of GTR: Gaussian Splatting Tracking and Reconstruction of Unknown Objects Based on Appearance and Geometric Complexity." pith.science (2026). https://pith.science/paper/VENISOM6

@misc{pith2026250511905,
  author       = {Pith},
  title        = {Pith review of: GTR: Gaussian Splatting Tracking and Reconstruction of Unknown Objects Based on Appearance and Geometric Complexity},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/VENISOM6}},
  note         = {Machine review of arXiv:2505.11905}
}
read the original abstract

We present a novel method for 6-DoF object tracking and high-quality 3D reconstruction from monocular RGBD video. Existing methods, while achieving impressive results, often struggle with complex objects, particularly those exhibiting symmetry, intricate geometry or complex appearance. To bridge these gaps, we introduce an adaptive method that combines 3D Gaussian Splatting, hybrid geometry/appearance tracking, and key frame selection to achieve robust tracking and accurate reconstructions across a diverse range of objects. Additionally, we present a benchmark covering these challenging object classes, providing high-quality annotations for evaluating both tracking and reconstruction performance. Our approach demonstrates strong capabilities in recovering high-fidelity object meshes, setting a new standard for single-sensor 3D reconstruction in open-world environments.

Figures

Figures reproduced from arXiv: 2505.11905 by the authors.

Figure 1
Figure 1. We present GTR, an adaptive method for 6-DoF object tracking and 3D reconstruction from monocular RGBD video. Although [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. Pipeline Flow: Our method processes sequential RGB-D frames and estimates coarse relative poses using the (1) Keypoint Detection & Tracking module. If the poses meet the visibility criteria, the frames are added to the keyframe pool (2). To refine the keyframe poses, the (3) Pose Refinement module—incorporating ICP, 3DGS-based render-and-compare, and pose graph optimization techniques—is applied. Finally, shapes are… view at source ↗
Figure 3
Figure 3. Appearance and Geometric Complexity: The de￾tected keypoints from SIFT are visualized in the left figure. The appearance complexity can vary depending on the viewpoint, even for the same object. Lower complexity makes keypoint tracking more difficult. We also visualize the geometric complexity at each pixel using a color map based on our proposed method. Higher complexity makes the registration problem easier. for f… view at source ↗
Figures from the paper (22 more)
Figure 4
Figure 4. Figure 4: Coarse Pose Solver: We use TEASER++ as part of our keypoint detection and tracking module to obtain an initial pose estimate for the new frame relative to the last keyframe using coarse correspondences. 3 [PITH_FULL_IMAGE:figures/full_fig_p003_4.png]
Figure 6
Figure 6. Figure 6: 3DGS Pose Refiner: We use 3DGS to improve the initial pose estimate using render-and-compare approach. Geometric Complexity To estimate the geometric com￾plexity of the current frame, we compute the mean local curvature of all visible points. The curvature at each poin…
Figure 7
Figure 7. Figure 7: GTR3D Benchmark: Our benchmark includes six objects designed to address some of the most challenging issues faced by existing trackers, such as axis-symmetry, high-frequency geometric details, and diverse materials. To evaluate tracking and reconstruction performance u…
Figure 8
Figure 8. Figure 8: Qualitative results on GTR3D Synthetic: Our method shows overall best performance across baselines resulting in better object reconstructions. Here red shows ground-truth trajectory and blue shows predicted. We visualize the reconstructed 3D mesh for all approaches [P…
Figure 9
Figure 9. Figure 9: Qualitative results on GTR3D Real: Our method demonstrates consistent performance for different types of objects resulting in better object reconstructions on average. Here red shows ground-truth trajectory and blue shows predicted. We visualize the reconstructed 3D me…
Figure 11
Figure 11. Figure 11: Object Trajectory in GTR3D Real B. Pose Annotations of GTR3D Real Dataset To evaluate tracking performance, pose annotations are re￾quired for the real data. For this, we first created 3D models with the help of digital artists. Using the created 3D models, coarse pos…
Figure 12
Figure 12. Figure 12: 6D Pose Annotation 1 [PITH_FULL_IMAGE:figures/full_fig_p011_12.png]
Figure 13
Figure 13. Figure 13: Tracking and Reconstruction on GTR3D Real [PITH_FULL_IMAGE:figures/full_fig_p012_13.png]
Figure 14
Figure 14. Figure 14: Execution Time Comparison E. Reconstruction Results We show our reconstruction results on GTR3D Real from both appearance and geometry perspectives, comparing them with 2 SOTA baselines: BundleSDF [44], Gaussian￾SLAM [50]. In the [PITH_FULL_IMAGE:figures/full_fig_p01…
Figure 15
Figure 15. Figure 15: Our 3DGS Reconstructions on GTR3D Real 4 [PITH_FULL_IMAGE:figures/full_fig_p014_15.png]
Figure 16
Figure 16. Figure 16: 3DGS Reconstruction Comparison on GTR3D Real, Box Object [PITH_FULL_IMAGE:figures/full_fig_p015_16.png]
Figure 17
Figure 17. Figure 17: 3DGS Reconstruction Comparison on GTR3D Real, Can Object [PITH_FULL_IMAGE:figures/full_fig_p015_17.png]
Figure 18
Figure 18. Figure 18: 3DGS Reconstruction Comparison on GTR3D Real, Dino Object [PITH_FULL_IMAGE:figures/full_fig_p016_18.png]
Figure 19
Figure 19. Figure 19: 3DGS Reconstruction Comparison on GTR3D Real, Bottle Object [PITH_FULL_IMAGE:figures/full_fig_p016_19.png]
Figure 20
Figure 20. Figure 20: 3DGS Reconstruction Comparison on GTR3D Real, Spray Object [PITH_FULL_IMAGE:figures/full_fig_p017_20.png]
Figure 21
Figure 21. Figure 21: 3DGS Reconstruction Comparison on GTR3D Real, Drill Object [PITH_FULL_IMAGE:figures/full_fig_p017_21.png]
Figure 22
Figure 22. Figure 22: Mesh Reconstruction Comparison on GTR3D Real, Box Object [PITH_FULL_IMAGE:figures/full_fig_p018_22.png]
Figure 23
Figure 23. Figure 23: Mesh Reconstruction Comparison on GTR3D Real, Can Object [PITH_FULL_IMAGE:figures/full_fig_p018_23.png]
Figure 24
Figure 24. Figure 24: Mesh Reconstruction Comparison on GTR3D Real, Dino Object [PITH_FULL_IMAGE:figures/full_fig_p019_24.png]
Figure 25
Figure 25. Figure 25: Mesh Reconstruction Comparison on GTR3D Real, Bottle Object [PITH_FULL_IMAGE:figures/full_fig_p019_25.png]
Figure 26
Figure 26. Figure 26: Mesh Reconstruction Comparison on GTR3D Real, Spray Object [PITH_FULL_IMAGE:figures/full_fig_p020_26.png]
Figure 27
Figure 27. Figure 27: Mesh Reconstruction Comparison on GTR3D Real, Drill Object [PITH_FULL_IMAGE:figures/full_fig_p020_27.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

55 extracted references · 49 canonical work pages

  1. [1]

    Scene coordinate reconstruction: Pos- ing of image collections via incremental learning of a relo- calizer

    Eric Brachmann, Jamie Wynn, Shuai Chen, Tommaso Cav- allari, ´Aron Monszpart, Daniyar Turmukhambetov, and Vic- tor Adrian Prisacariu. Scene coordinate reconstruction: Pos- ing of image collections via incremental learning of a relo- calizer. In ECCV, 2025. 6

    work page 2025

  2. [2]

    Freeze: Training-free zero-shot 6d pose estimation with geometric and vision foundation models

    Andrea Caraffa, Davide Boscaini, Amir Hamza, and Fabio Poiesi. Freeze: Training-free zero-shot 6d pose estimation with geometric and vision foundation models. In ECCV,

    work page

  3. [3]

    Learn- ing canonical shape space for category-level 6d object pose and size estimation

    Dengsheng Chen, Jun Li, Zheng Wang, and Kai Xu. Learn- ing canonical shape space for category-level 6d object pose and size estimation. In CVPR, 2020. 2

    work page 2020

  4. [4]

    Bootstap: Boot- strapped training for tracking-any-point

    Carl Doersch, Pauline Luc, Yi Yang, Dilara Gokay, Skanda Koppula, Ankush Gupta, Joseph Heyward, Ignacio Rocco, Ross Goroshin, Jo ˜ao Carreira, et al. Bootstap: Boot- strapped training for tracking-any-point. arXiv preprint arXiv:2402.00847, 2024. 3, 2

    arXiv 2024

  5. [5]

    Dynamic slam: The need for speed

    Mina Henein, Jun Zhang, Robert Mahony, and Viorela Ila. Dynamic slam: The need for speed. In ICRA, 2020. 2

    work page 2020

  6. [6]

    Epos: esti- mating 6d pose of objects with symmetries

    Tomas Hodan, Daniel Barath, and Jiri Matas. Epos: esti- mating 6d pose of objects with symmetries. In CVPR, 2020. 2

    work page 2020

  7. [7]

    Zeroshape: Regression-based zero-shot shape reconstruction

    Zixuan Huang, Stefan Stojanov, Anh Thai, Varun Jampani, and James M Rehg. Zeroshape: Regression-based zero-shot shape reconstruction. In CVPR, 2024. 2

    work page 2024

  8. [8]

    Diffusionnocs: Managing sym- metry and uncertainty in sim2real multi-modal category- level pose estimation

    Takuya Ikeda, Sergey Zakharov, Tianyi Ko, Muham- mad Zubair Irshad, Robert Lee, Katherine Liu, Rares Am- brus, and Koichi Nishiwaki. Diffusionnocs: Managing sym- metry and uncertainty in sim2real multi-modal category- level pose estimation. In IROS, 2024. 2

    work page 2024

Show all 55 references
  1. [9]

    Centersnap: Single-shot multi- object 3d shape reconstruction and categorical 6d pose and size estimation

    Muhammad Zubair Irshad, Thomas Kollar, Michael Laskey, Kevin Stone, and Zsolt Kira. Centersnap: Single-shot multi- object 3d shape reconstruction and categorical 6d pose and size estimation. 2022. 2

    2022

  2. [10]

    Shapo: Im- 8 plicit representations for multi-object shape appearance and pose optimization

    Muhammad Zubair Irshad, Sergey Zakharov, Rares Ambrus, Thomas Kollar, Zsolt Kira, and Adrien Gaidon. Shapo: Im- 8 plicit representations for multi-object shape appearance and pose optimization. 2022. 2

    2022

  3. [11]

    Neural fields in robotics: A survey,

    Muhammad Zubair Irshad, Mauro Comi, Yen-Chen Lin, Nick Heppert, Abhinav Valada, Rares Ambrus, Zsolt Kira, and Jonathan Tremblay. Neural fields in robotics: A survey,

  4. [12]

    Zero-shot multi-object scene completion

    Shun Iwase, Katherine Liu, Vitor Guizilini, Adrien Gaidon, Kris Kitani, Rares, Ambrus,, and Sergey Zakharov. Zero-shot multi-object scene completion. In ECCV, 2024. 2

    2024

  5. [13]

    Splatam: Splat, track & map 3d gaussians for dense rgb-d slam

    Nikhil Keetha, Jay Karhade, Krishna Murthy Jatavallabhula, Gengshan Yang, Sebastian Scherer, Deva Ramanan, and Jonathon Luiten. Splatam: Splat, track & map 3d gaussians for dense rgb-d slam. In CVPR, 2024. 2

    2024

  6. [14]

    Deep learning of local rgb-d patches for 3d object detection and 6d pose estimation

    Wadim Kehl, Fausto Milletari, Federico Tombari, Slobodan Ilic, and Nassir Navab. Deep learning of local rgb-d patches for 3d object detection and 6d pose estimation. In ECCV,

  7. [15]

    Ssd-6d: Making rgb-based 3d de- tection and 6d pose estimation great again

    Wadim Kehl, Fabian Manhardt, Federico Tombari, Slobodan Ilic, and Nassir Navab. Ssd-6d: Making rgb-based 3d de- tection and 6d pose estimation great again. In ICCV, 2017. 2

    2017

  8. [16]

    3d gaussian splatting for real-time radiance field rendering

    Bernhard Kerbl, Georgios Kopanas, Thomas Leimk ¨uhler, and George Drettakis. 3d gaussian splatting for real-time radiance field rendering. ACM Trans. Graph., 2023. 2

    2023

  9. [17]

    vmap: Vectorised object mapping for neural field slam

    Xin Kong, Shikun Liu, Marwan Taher, and Andrew J Davi- son. vmap: Vectorised object mapping for neural field slam. In CVPR, 2023. 2

    2023

  10. [18]

    Jiahui Lei, Srinath Sridhar, Paul Guerrero, Minhyuk Sung, Niloy Mitra, and Leonidas J. Guibas. Pix2surf: Learning parametric 3d surface models of objects from images. In ECCV, 2020. 2

    2020

  11. [19]

    Fsd: Fast self-supervised single rgb-d to categorical 3d objects

    Mayank Lunayach, Sergey Zakharov, Dian Chen, Rares Am- brus, Zsolt Kira, and Muhammad Zubair Irshad. Fsd: Fast self-supervised single rgb-d to categorical 3d objects. In ICRA, 2024. 2

    2024

  12. [20]

    Hidenobu Matsuki, Riku Murai, Paul H. J. Kelly, and An- drew J. Davison. Gaussian Splatting SLAM. InCVPR, 2024. 2

    2024

  13. [21]

    Extracting Triangular 3D Models, Materials, and Light- ing From Images

    Jacob Munkberg, Jon Hasselgren, Tianchang Shen, Jun Gao, Wenzheng Chen, Alex Evans, Thomas M¨uller, and Sanja Fi- dler. Extracting Triangular 3D Models, Materials, and Light- ing From Images. In CVPR, 2022. 2

    2022

  14. [22]

    Tard ´os

    Ra ´ul Mur-Artal and Juan D. Tard ´os. ORB-SLAM2: an open-source SLAM system for monocular, stereo and RGB- D cameras. T-RO, 2017. 2

    2017

  15. [23]

    Unisurf: Unifying neural implicit surfaces and radiance fields for multi-view reconstruction

    Michael Oechsle, Songyou Peng, and Andreas Geiger. Unisurf: Unifying neural implicit surfaces and radiance fields for multi-view reconstruction. In ICCV, 2021. 2

    2021

  16. [24]

    Found- pose: Unseen object pose estimation with foundation fea- tures

    Evin Pınar ¨Ornek, Yann Labb ´e, Bugra Tekin, Lingni Ma, Cem Keskin, Christian Forster, and Tomas Hodan. Found- pose: Unseen object pose estimation with foundation fea- tures. In ECCV, 2025. 2

    2025

  17. [25]

    Bb8: A scalable, accurate, robust to partial occlusion method for predicting the 3d poses of challenging objects without using depth

    Mahdi Rad and Vincent Lepetit. Bb8: A scalable, accurate, robust to partial occlusion method for predicting the 3d poses of challenging objects without using depth. In ICCV, 2017. 2

    2017

  18. [26]

    Sam 2: Segment anything in images and videos

    Nikhila Ravi, Valentin Gabeur, Yuan-Ting Hu, Ronghang Hu, Chaitanya Ryali, Tengyu Ma, Haitham Khedr, Roman R¨adle, Chloe Rolland, Laura Gustafson, Eric Mintun, Junt- ing Pan, Kalyan Vasudev Alwala, Nicolas Carion, Chao- Yuan Wu, Ross Girshick, Piotr Doll´ar, and Christoph Feic...

    2024 arXiv

  19. [27]

    Meshsdf: Differentiable iso-surface extraction

    Edoardo Remelli, Artem Lukoianov, Stephan Richter, Benoit Guillard, Timur Bagautdinov, Pierre Baque, and Pascal Fua. Meshsdf: Differentiable iso-surface extraction. In NeurIPS,

  20. [28]

    Efficient variants of the icp algorithm

    Szymon Rusinkiewicz and Marc Levoy. Efficient variants of the icp algorithm. In 3DIM, 2001. 4

    2001

  21. [29]

    A learned stereo depth system for robotic manipulation in homes

    Krishna Shankar, Mark Tjersland, Jeremy Ma, Kevin Stone, and Max Bajracharya. A learned stereo depth system for robotic manipulation in homes. RA-L, 2022. 1

    2022

  22. [30]

    Flowmap: High-quality camera poses, intrinsics, and depth via gradient descent

    Cameron Smith, David Charatan, Ayush Tewari, and Vincent Sitzmann. Flowmap: High-quality camera poses, intrinsics, and depth via gradient descent. In 3DV, 2025. 6, 3

    2025

  23. [31]

    Rechert, Florian Steidle, Dongheui Lee, and Rudolph Triebel

    Manuel Stoiber, Mariam Elsayed, Anne E. Rechert, Florian Steidle, Dongheui Lee, and Rudolph Triebel. Fusing vi- sual appearance and geometry for multi-modality 6dof object tracking. In IROS, 2023. 6, 1

    2023

  24. [32]

    Loftr: Detector-free local feature matching with transformers

    Jiaming Sun, Zehong Shen, Yuang Wang, Hujun Bao, and Xiaowei Zhou. Loftr: Detector-free local feature matching with transformers. In CVPR, 2021. 3, 2

    2021

  25. [33]

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

    Martin Sundermeyer, Zoltan-Csaba Marton, Maximilian Durner, Manuel Brucker, and Rudolph Triebel. Implicit 3d orientation learning for 6d object detection from rgb images. In ECCV, 2018. 2

    2018

  26. [34]

    Raft: Recurrent all-pairs field transforms for optical flow

    Zachary Teed and Jia Deng. Raft: Recurrent all-pairs field transforms for optical flow. In ECCV, 2020. 3

    2020

  27. [35]

    DROID-SLAM: Deep Vi- sual SLAM for Monocular, Stereo, and RGB-D Cameras

    Zachary Teed and Jia Deng. DROID-SLAM: Deep Vi- sual SLAM for Monocular, Stereo, and RGB-D Cameras. NeurIPS, 2021. 2

    2021

  28. [36]

    Latent-class hough forests for 3d object de- tection and pose estimation

    Alykhan Tejani, Danhang Tang, Rigas Kouskouridas, and Tae-Kyun Kim. Latent-class hough forests for 3d object de- tection and pose estimation. In ECCV, 2014. 2

    2014

  29. [37]

    Real-time seamless single shot 6d object pose prediction

    Bugra Tekin, Sudipta N Sinha, and Pascal Fua. Real-time seamless single shot 6d object pose prediction. In CVPR,

  30. [38]

    Shape prior deformation for categorical 6d object pose and size estima- tion

    Meng Tian, Marcelo H Ang, and Gim Hee Lee. Shape prior deformation for categorical 6d object pose and size estima- tion. In ECCV, 2020. 2

    2020

  31. [39]

    Densefusion: 6d object pose estimation by iterative dense fusion

    Chen Wang, Danfei Xu, Yuke Zhu, Roberto Mart ´ın-Mart´ın, Cewu Lu, Li Fei-Fei, and Silvio Savarese. Densefusion: 6d object pose estimation by iterative dense fusion. In CVPR,

  32. [40]

    PyPose: A library for robot learning with physics-based optimization

    Chen Wang, Dasong Gao, Kuan Xu, Junyi Geng, Yaoyu Hu, Yuheng Qiu, Bowen Li, Fan Yang, Brady Moon, Abhinav Pandey, Aryan, Jiahe Xu, Tianhao Wu, Haonan He, Daning Huang, Zhongqiang Ren, Shibo Zhao, Taimeng Fu, Pranay Reddy, Xiao Lin, Wenshan Wang, Jingnan Shi, Rajat Talak, Kun C...

    2023

  33. [41]

    Neus: Learning neural implicit surfaces by volume rendering for multi-view reconstruction

    Peng Wang, Lingjie Liu, Yuan Liu, Christian Theobalt, Taku Komura, and Wenping Wang. Neus: Learning neural implicit surfaces by volume rendering for multi-view reconstruction. In NeurIPS, 2021. 2

    2021

  34. [42]

    Image quality assessment: from error visibility to structural similarity

    Zhou Wang, Alan C Bovik, Hamid R Sheikh, and Eero P Simoncelli. Image quality assessment: from error visibility to structural similarity. TIP, 2004. 5

    2004

  35. [43]

    Bundletrack: 6d pose track- ing for novel objects without instance or category-level 3d models

    B Wen and Kostas E Bekris. Bundletrack: 6d pose track- ing for novel objects without instance or category-level 3d models. In IROS, 2021. 2, 6

    2021

  36. [44]

    Bundlesdf: Neural 6-dof tracking and 3d reconstruction of unknown objects

    Bowen Wen, Jonathan Tremblay, Valts Blukis, Stephen Tyree, Thomas M ¨uller, Alex Evans, Dieter Fox, Jan Kautz, and Stan Birchfield. Bundlesdf: Neural 6-dof tracking and 3d reconstruction of unknown objects. In CVPR, 2023. 1, 2, 6, 7

    2023

  37. [45]

    Foundationpose: Unified 6d pose estimation and tracking of novel objects

    Bowen Wen, Wei Yang, Jan Kautz, and Stan Birchfield. Foundationpose: Unified 6d pose estimation and tracking of novel objects. In CVPR, 2024. 1, 2, 6

    2024

  38. [46]

    Object- compositional neural implicit surfaces

    Qianyi Wu, Xian Liu, Yuedong Chen, Kejie Li, Chuanxia Zheng, Jianfei Cai, and Jianmin Zheng. Object- compositional neural implicit surfaces. In ECCV, 2022. 2

    2022

  39. [47]

    Posecnn: A convolutional neural network for 6d object pose estimation in cluttered scenes

    Yu Xiang, Tanner Schmidt, Venkatraman Narayanan, and Dieter Fox. Posecnn: A convolutional neural network for 6d object pose estimation in cluttered scenes. In RSS, 2018. 2

    2018

  40. [48]

    Learning object-compositional neural radiance field for ed- itable scene rendering

    Bangbang Yang, Yinda Zhang, Yinghao Xu, Yijin Li, Han Zhou, Hujun Bao, Guofeng Zhang, and Zhaopeng Cui. Learning object-compositional neural radiance field for ed- itable scene rendering. In ICCV, 2021. 2

    2021

  41. [49]

    H. Yang, J. Shi, and L. Carlone. TEASER: Fast and Certifi- able Point Cloud Registration. T-RO, 2020. 3, 2

    2020

  42. [50]

    Gaussian-slam: Photo-realistic dense slam with gaus- sian splatting

    Vladimir Yugay, Yue Li, Theo Gevers, and Martin R Os- wald. Gaussian-slam: Photo-realistic dense slam with gaus- sian splatting. arXiv preprint arXiv:2312.10070, 2023. 4, 6, 7, 2, 3

    2023 arXiv

  43. [51]

    Dpod: Dense 6d pose object detector in rgb images

    Sergey Zakharov, Ivan Shugurov, and Slobodan Ilic. Dpod: Dense 6d pose object detector in rgb images. InICCV, 2019. 2

    2019

  44. [52]

    Generative category-level object pose estimation via diffusion models

    Jiyao Zhang, Mingdong Wu, and Hao Dong. Generative category-level object pose estimation via diffusion models. NeurIPS, 2024. 2

    2024

  45. [53]

    Cameras as rays: Pose estimation via ray diffusion

    Jason Y Zhang, Amy Lin, Moneish Kumar, Tzu-Hsuan Yang, Deva Ramanan, and Shubham Tulsiani. Cameras as rays: Pose estimation via ray diffusion. In ICLR, 2024. 6

    2024

  46. [54]

    Oswald, and Marc Polle- feys

    Zihan Zhu, Songyou Peng, Viktor Larsson, Weiwei Xu, Hu- jun Bao, Zhaopeng Cui, Martin R. Oswald, and Marc Polle- feys. Nice-slam: Neural implicit scalable encoding for slam. In CVPR, 2022. 2 10 GTR: Gaussian Splatting Tracking and Reconstruction of Unknown Objects Based on App...

    2022

  47. [55]

    Among these, 3DGS reconstruction and PGO are the two most time-consuming components

    (0.048s), TEASER++ [49] (0.014s), ICP (0.077s), 3DGS reconstruction and render-and-compare (29.471s), and PGO (7959.463s). Among these, 3DGS reconstruction and PGO are the two most time-consuming components. For the non-linear optimization of PGO, we employed a simple implemen...

Pith tools

Reviewed August 15, 2026 · model on record in the stance chip above.