pith. machine review for the scientific record. sign in

arxiv: 2605.14393 · v1 · submitted 2026-05-14 · 💻 cs.CV

Recognition: 2 theorem links

· Lean Theorem

Analogical Trajectory Transfer

Junho Kim , Eun Sun Lee , Gwangtak Bae , Seunggu Kang , Young Min Kim
Authors on Pith no claims yet

Pith reviewed 2026-05-15 01:41 UTC · model grok-4.3

classification 💻 cs.CV
keywords analogical trajectory transferobject-centric clusteringcross-scene mapping3D foundation modelssmooth map predictionmotion transferspatial coherencerobotics applications
0
0 comments X

The pith

Scenes are partitioned into object-centric clusters whose cross-scene mappings are predicted hierarchically from 3D foundation features and then assembled and refined to transfer trajectories while preserving semantics and avoiding clashes.

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

The paper sets out to translate motion trajectories from one 3D environment to a functionally analogous place in a second environment whose layout, scale, and object placement may differ. Direct semantic matching often produces collisions or torn paths, so the work decomposes the task into spatially separated subproblems. Each subproblem is solved by predicting a smooth mapping over object-centric clusters whose contextual features come from an off-the-shelf 3D foundation model. The per-cluster maps are combined combinatorially and then locally corrected, yielding a single coherent trajectory that requires no scene-specific training and runs in roughly half a second.

Core claim

By partitioning each scene into object-centric clusters, estimating hierarchical smooth cross-scene mappings from 3D foundation-model features that capture both object and open-space context, combinatorially assembling the per-cluster maps, and refining the composite result to eliminate collisions and distortions, a semantically consistent and spatially coherent trajectory transfer is obtained without any per-scene training.

What carries the argument

Hierarchical smooth map prediction over object-centric clusters, using 3D foundation-model features to supply contextual information about object arrangements and open spaces, followed by combinatorial assembly and collision-refinement.

If this is right

  • Trajectory transfer completes in approximately 0.6 seconds without any scene-specific training or manual tuning.
  • The resulting transfers outperform baselines that rely on LLMs, VLMs, or scene-graph matching on both semantic fidelity and geometric coherence.
  • The same pipeline directly supports virtual co-presence, multi-trajectory transfer, camera-path transfer, and human-to-robot motion transfer.

Where Pith is reading between the lines

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

  • Because the method never trains on the target scenes, it could be applied on the fly to newly scanned or procedurally generated environments.
  • The same cluster-wise decomposition might be reused to transfer other continuous quantities such as force fields or affordance maps.
  • If stronger 3D foundation models become available, the same pipeline would immediately improve without any change to the assembly or refinement stages.

Load-bearing premise

The 3D foundation model features already encode enough information about object and open-space arrangements to produce accurate cross-scene mappings that keep both semantics and functionality intact.

What would settle it

A pair of scenes whose functional analogy is clear to humans yet whose 3D foundation features produce a transferred trajectory that intersects obstacles or violates functional constraints after refinement.

Figures

Figures reproduced from arXiv: 2605.14393 by Eun Sun Lee, Gwangtak Bae, Junho Kim, Seunggu Kang, Young Min Kim.

Figure 1
Figure 1. Figure 1: The goal of analogical trajectory transfer is to map trajectory points from one scene to semantically and spatially analogous locations in another, enabling applications in order-critical, multi-object interaction tasks such as sim-to-real robot motion transfer and camera trajectory transfer. understanding rich spatial context, as trajectories are shaped not only by individual objects but by their arrangem… view at source ↗
Figure 2
Figure 2. Figure 2: Overview of our approach. We first match object clusters (node colors represent cluster [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Qualitative comparison of analogical trajectory transfer on various datasets and baselines. [PITH_FULL_IMAGE:figures/full_fig_p008_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Visualizations of applications enabled by analogical trajectory transfer. [PITH_FULL_IMAGE:figures/full_fig_p009_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Visualization of top-L trajec￾tory transfer candidates in 3D-FRONT. Virtual Co-presence. Given trajectories in multiple differ￾ent scenes, our method can transfer all of them into a single shared reference scene. This enables virtual co-presence scenarios where participants from separate physical envi￾ronments aim to share a common virtual space [Keshavarzi et al., 2022]. Camera Trajectory Transfer. Our me… view at source ↗
Figure 6
Figure 6. Figure 6: Graphical user interface (GUI) used for paired (single- and multiple ground-truth) annota [PITH_FULL_IMAGE:figures/full_fig_p016_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: Additional qualitative results on 3D-FRONT scenes. Each row shows the target scene [PITH_FULL_IMAGE:figures/full_fig_p018_7.png] view at source ↗
Figure 8
Figure 8. Figure 8: Additional qualitative results on ARKitScenes (top two rows) and cross-domain pairs [PITH_FULL_IMAGE:figures/full_fig_p019_8.png] view at source ↗
read the original abstract

We study analogical trajectory transfer, where the goal is to translate motion trajectories in one 3D environment to a semantically analogous location in another. Such a capacity would enable machines to perform analogical spatial reasoning, with applications in AR/VR co-presence, content creation, and robotics. However, even semantically similar scenes can still differ substantially in object placement, scale, and layout, so naively matching semantics leads to collisions or geometric distortions. Furthermore, finding where each trajectory point should transfer to has a large search space, as the mapping must preserve semantics and functionality without tearing the trajectory apart or causing collisions. Our key insight is to decompose the problem into spatially segregated subproblems and merge their solutions to produce semantically consistent and spatially coherent transfers. Specifically, we partition scenes into object-centric clusters and estimate cross-scene mappings via hierarchical smooth map prediction, using 3D foundation model features that encode contextual information from object and open-space arrangements. We then combinatorially assemble the per-cluster maps into an initial transfer and refine the result to remove collisions and distortions, yielding a spatially coherent trajectory. Our method does not require training, attains a fast runtime around 0.6 seconds, and outperforms baselines based on LLMs, VLMs, and scene graph matching. We further showcase applications in virtual co-presence, multi-trajectory transfer, camera transfer, and human-to-robot motion transfer, which indicates the broad applicability of our work to AR/VR and robotics.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 2 minor

Summary. The manuscript proposes a training-free approach to analogical trajectory transfer between semantically similar 3D scenes that may differ in layout, scale, and object placement. Scenes are partitioned into object-centric clusters; cross-scene mappings are estimated via hierarchical smooth map prediction that leverages 3D foundation model features encoding object and open-space context. Per-cluster maps are combinatorially assembled into an initial transfer and then refined to remove collisions and distortions, producing a spatially coherent result. The method is reported to run in ~0.6 s and to outperform LLM-, VLM-, and scene-graph-based baselines, with applications demonstrated in virtual co-presence, multi-trajectory transfer, camera transfer, and human-to-robot motion transfer.

Significance. If the experimental claims are substantiated, the work would offer a practical, training-free solution to a challenging problem in spatial analogical reasoning, with clear utility for AR/VR content creation and robotics. The decomposition strategy and reliance on off-the-shelf 3D foundation models are attractive because they avoid scene-specific training. The reported speed and breadth of applications strengthen the potential impact in computer vision and robotics venues.

major comments (2)
  1. [Abstract] Abstract and experimental section: the central claim of outperformance over LLM, VLM, and scene-graph baselines is stated without any quantitative tables, error metrics, ablation studies, or failure-case analysis visible in the manuscript. Because the contribution rests on these empirical results, their absence is load-bearing and prevents verification of the method's superiority.
  2. [Method] Method description (hierarchical smooth map prediction paragraph): the claim that 3D foundation model features supply 'contextual information from object and open-space arrangements' sufficient to produce functionality-preserving mappings is asserted without supporting analysis or counter-examples. No evidence is given that the embeddings reliably distinguish geometrically similar but functionally distinct configurations (e.g., narrow corridor vs. wide aisle), which directly affects the correctness of the subsequent combinatorial assembly and refinement steps.
minor comments (2)
  1. The runtime figure of 0.6 seconds is given without hardware specification, per-component timing breakdown, or comparison to baseline runtimes.
  2. Notation for the hierarchical map prediction and combinatorial assembly steps could be clarified with a concise pseudocode listing or explicit variable definitions.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive and detailed review. The comments highlight important areas where additional empirical support and analysis will strengthen the manuscript. We address each major comment below and will revise the paper accordingly.

read point-by-point responses

  1. Referee: [Abstract] Abstract and experimental section: the central claim of outperformance over LLM, VLM, and scene-graph baselines is stated without any quantitative tables, error metrics, ablation studies, or failure-case analysis visible in the manuscript. Because the contribution rests on these empirical results, their absence is load-bearing and prevents verification of the method's superiority.

    Authors: We agree that the empirical claims require explicit quantitative support for verification. The submitted manuscript emphasized the method and qualitative demonstrations, but the revised version will expand the experimental section with quantitative tables reporting metrics including average trajectory error, collision rates, and functional preservation scores against the LLM, VLM, and scene-graph baselines. Ablation studies isolating the hierarchical mapping, combinatorial assembly, and refinement components, plus a dedicated failure-case analysis, will be added. revision: yes

  2. Referee: [Method] Method description (hierarchical smooth map prediction paragraph): the claim that 3D foundation model features supply 'contextual information from object and open-space arrangements' sufficient to produce functionality-preserving mappings is asserted without supporting analysis or counter-examples. No evidence is given that the embeddings reliably distinguish geometrically similar but functionally distinct configurations (e.g., narrow corridor vs. wide aisle), which directly affects the correctness of the subsequent combinatorial assembly and refinement steps.

    Authors: We thank the referee for this precise observation. While the foundation models are pretrained on diverse 3D data and their features are intended to capture contextual arrangements, we acknowledge the absence of direct supporting analysis. The revised manuscript will augment the method section with feature similarity visualizations and quantitative comparisons for geometrically similar yet functionally distinct pairs (e.g., narrow corridors vs. wide aisles). We will also discuss how these distinctions propagate to the combinatorial assembly and refinement stages, including counter-examples illustrating both successful mappings and cases corrected by refinement. revision: yes

Circularity Check

0 steps flagged

No circularity: derivation relies on external 3D foundation models and independent geometric steps

full rationale

The paper's chain decomposes the transfer task into clustering, hierarchical mapping via pretrained 3D features, combinatorial assembly, and collision-refinement. None of these steps are shown to reduce to fitted parameters renamed as predictions, self-definitional equations, or load-bearing self-citations. The method explicitly states it requires no training and uses off-the-shelf foundation models whose embeddings are treated as independent inputs. No uniqueness theorems or ansatzes are smuggled via author citations, and no known empirical pattern is merely renamed. The central claim therefore remains non-circular by the paper's own presentation.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

The central claim rests on the assumption that foundation-model features already encode the necessary semantic and spatial context for cross-scene mapping; no new entities are postulated and no parameters are fitted inside the method itself.

axioms (1)
  • domain assumption 3D foundation model features encode contextual information from object and open-space arrangements sufficient for semantically consistent mappings
    Invoked in the description of hierarchical smooth map prediction

pith-pipeline@v0.9.0 · 5563 in / 1149 out tokens · 22023 ms · 2026-05-15T01:41:07.075818+00:00 · methodology

discussion (0)

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

Lean theorems connected to this paper

Citations machine-checked in the Pith Canon. Every link opens the source theorem in the public Lean library.

What do these tags mean?
matches
The paper's claim is directly supported by a theorem in the formal canon.
supports
The theorem supports part of the paper's argument, but the paper may add assumptions or extra steps.
extends
The paper goes beyond the formal theorem; the theorem is a base layer rather than the whole result.
uses
The paper appears to rely on the theorem as machinery.
contradicts
The paper's claim conflicts with a theorem or certificate in the canon.
unclear
Pith found a possible connection, but the passage is too broad, indirect, or ambiguous to say the theorem truly supports the claim.

Reference graph

Works this paper leans on

192 extracted references · 192 canonical work pages

  1. [1]

    Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV) , month =

    Kim, Junho and Choi, Changwoon and Jang, Hojun and Kim, Young Min , title =. Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV) , month =. 2021 , pages =

    work page 2021

  2. [2]

    The Farthest Point Strategy for Progressive Image Sampling , volume =

    Eldar, Yuval and Lindenbaum, Michael and Porat, Moshe and Zeevi, Yehoshua , year =. The Farthest Point Strategy for Progressive Image Sampling , volume =. IEEE Transactions on Image Processing , doi =

    work page

  3. [3]

    , title =

    Eddy, William F. , title =. 1977 , issue_date =. doi:10.1145/355759.355766 , journal =

    work page doi:10.1145/355759.355766 1977

  4. [4]

    Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV) , year=

    3d-front: 3d furnished rooms with layouts and semantics , author=. Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV) , year=

    work page

  5. [5]

    International Journal of Computer Vision (IJCV) , pages=

    3d-future: 3d furniture shape with texture , author=. International Journal of Computer Vision (IJCV) , pages=

    work page

  6. [6]

    Acta Psychologica , volume=

    Why do children struggle on analogical reasoning tasks? Considering the role of problem format by measuring visual attention , author=. Acta Psychologica , volume=. 2022 , publisher=

    work page 2022

  7. [7]

    and Kaiser,

    Vaswani, Ashish and Shazeer, Noam and Parmar, Niki and Uszkoreit, Jakob and Jones, Llion and Gomez, Aidan N. and Kaiser,. Attention is all you need , year =

    work page

  8. [8]

    BERT: Pre-training of deep bidirectional transformers for language understanding

    Devlin, Jacob and Chang, Ming-Wei and Lee, Kenton and Toutanova, Kristina. BERT : Pre-training of Deep Bidirectional Transformers for Language Understanding. Proceedings of the Conference of the North A merican Chapter of the Association for Computational Linguistics (NAACL). 2019. doi:10.18653/v1/N19-1423

    work page doi:10.18653/v1/n19-1423 2019

  9. [9]

    Proceedings of the Conference on Robot Learning (CoRL) , year=

    Neural Attention Field: Emerging Point Relevance in 3D Scenes for One-Shot Dexterous Grasping , author=. Proceedings of the Conference on Robot Learning (CoRL) , year=

    work page

  10. [10]

    Proceedings of the International Conference on Learning Representations (ICLR) , year=

    An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale , author=. Proceedings of the International Conference on Learning Representations (ICLR) , year=

    work page

  11. [11]

    ArXiv , year=

    Representation Learning with Contrastive Predictive Coding , author=. ArXiv , year=

    work page

  12. [12]

    Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) , year=

    Exploring Simple Siamese Representation Learning , author=. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) , year=

    work page

  13. [13]

    Proceedings of the International Conference on Machine Learning (ICML) , year =

    A Simple Framework for Contrastive Learning of Visual Representations , author =. Proceedings of the International Conference on Machine Learning (ICML) , year =

    work page

  14. [14]

    Big Self-Supervised Models are Strong Semi-Supervised Learners , url =

    Chen, Ting and Kornblith, Simon and Swersky, Kevin and Norouzi, Mohammad and Hinton, Geoffrey E , booktitle =. Big Self-Supervised Models are Strong Semi-Supervised Learners , url =

    work page

  15. [15]

    and Leonidas Guibas and Or Litany

    Saining Xie and Jiatao Gu and Demi Guo and Qi, Charles R. and Leonidas Guibas and Or Litany. PointContrast: Unsupervised Pre-training for 3D Point Cloud Understanding. Proceedings of the European Conference on Computer Vision (ECCV). 2020

    work page 2020

  16. [16]

    IEEE Transactions on Pattern Analysis and Machine Intelligence (TPAMI) , volume=

    Convolutional hough matching networks for robust and efficient visual correspondence , author=. IEEE Transactions on Pattern Analysis and Machine Intelligence (TPAMI) , volume=. 2023 , publisher=

    work page 2023

  17. [17]

    Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) , year=

    Transformatcher: Match-to-match attention for semantic correspondence , author=. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) , year=

    work page

  18. [18]

    Proceedings of the European Conference on Computer Vision (ECCV) , year=

    Learning 3d keypoint descriptors for non-rigid shape matching , author=. Proceedings of the European Conference on Computer Vision (ECCV) , year=

    work page

  19. [19]

    Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) , year=

    Learning SO (3)-Invariant Semantic Correspondence via Local Shape Transform , author=. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) , year=

    work page

  20. [20]

    Computer Graphics Forum , volume=

    Neural semantic surface maps , author=. Computer Graphics Forum , volume=

    work page

  21. [21]

    ACM Transactions on Graphics (TOG) , volume=

    Blended intrinsic maps , author=. ACM Transactions on Graphics (TOG) , volume=. 2011 , publisher=

    work page 2011

  22. [22]

    Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) , year=

    Neural surface maps , author=. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) , year=

    work page

  23. [23]

    Proceedings of the European Conference on Computer Vision (ECCV) , year=

    Robo-abc: Affordance generalization beyond categories via semantic correspondence for robot manipulation , author=. Proceedings of the European Conference on Computer Vision (ECCV) , year=

    work page

  24. [24]

    Proceedings of the European Conference on Computer Vision (ECCV) , year=

    Guided semantic flow , author=. Proceedings of the European Conference on Computer Vision (ECCV) , year=

    work page

  25. [25]

    Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) , year=

    Fcss: Fully convolutional self-similarity for dense semantic correspondence , author=. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) , year=

    work page

  26. [26]

    Proceedings of the European Conference on Computer Vision (ECCV) , year=

    Parn: Pyramidal affine regression networks for dense semantic correspondence , author=. Proceedings of the European Conference on Computer Vision (ECCV) , year=

    work page

  27. [27]

    Proceedings of the Conference on Neural Information Processing Systems (NeurIPS) , volume=

    Recurrent transformer networks for semantic correspondence , author=. Proceedings of the Conference on Neural Information Processing Systems (NeurIPS) , volume=

    work page

  28. [28]

    Proceedings of the IEEE International Conference on Computer Vision (ICCV) , year=

    Dctm: Discrete-continuous transformation matching for semantic flow , author=. Proceedings of the IEEE International Conference on Computer Vision (ICCV) , year=

    work page

  29. [29]

    Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) , year=

    Misc210k: A large-scale dataset for multi-instance semantic correspondence , author=. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) , year=

    work page

  30. [30]

    Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV) , year=

    Discobox: Weakly supervised instance segmentation and semantic correspondence from box supervision , author=. Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV) , year=

    work page

  31. [31]

    Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) , year=

    Probing the 3d awareness of visual foundation models , author=. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) , year=

    work page

  32. [32]

    Proceedings of the Workshop on Language and Robotics at the Conference on Robot Learning (CoRL) , year=

    Clip-fields: Weakly supervised semantic fields for robotic memory , author=. Proceedings of the Workshop on Language and Robotics at the Conference on Robot Learning (CoRL) , year=

    work page

  33. [33]

    Advances in Neural Information Processing Systems , volume=

    Emergent correspondence from image diffusion , author=. Advances in Neural Information Processing Systems , volume=

    work page

  34. [34]

    Proceedings of the IEEE/CVF conference on computer vision and pattern recognition (CVPR) , year=

    Nice-slam: Neural implicit scalable encoding for slam , author=. Proceedings of the IEEE/CVF conference on computer vision and pattern recognition (CVPR) , year=

    work page

  35. [35]

    Communications of the ACM , volume=

    Nerf: Representing scenes as neural radiance fields for view synthesis , author=. Communications of the ACM , volume=. 2021 , publisher=

    work page 2021

  36. [36]

    Proceedings of the Conference on Neural Information Processing Systems (NeurIPS) , volume=

    Neural matching fields: Implicit representation of matching fields for visual correspondence , author=. Proceedings of the Conference on Neural Information Processing Systems (NeurIPS) , volume=

    work page

  37. [37]

    Proceedings of the IEEE International Conference on Robotics and Automation (ICRA) , year=

    nerf2nerf: Pairwise registration of neural radiance fields , author=. Proceedings of the IEEE International Conference on Robotics and Automation (ICRA) , year=

    work page

  38. [38]

    Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) , year=

    NeRF Analogies: Example-Based Visual Attribute Transfer for NeRFs , author=. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) , year=

    work page

  39. [39]

    Proceedings of the International Conference on Learning Representations (ICLR) , year=

    Diffusion Model for Dense Matching , author=. Proceedings of the International Conference on Learning Representations (ICLR) , year=

    work page

  40. [40]

    Proceedings of the IEEE International Conference on Robotics and Automation (ICRA) , year=

    Neural descriptor fields: Se (3)-equivariant object representations for manipulation , author=. Proceedings of the IEEE International Conference on Robotics and Automation (ICRA) , year=

    work page

  41. [41]

    Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) , month =

    Wei, Qiuhong Anna and Ding, Sijie and Park, Jeong Joon and Sajnani, Rahul and Poulenard, Adrien and Sridhar, Srinath and Guibas, Leonidas , title =. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) , month =

    work page

  42. [42]

    NeuS: Learning Neural Implicit Surfaces by Volume Rendering for Multi-view Reconstruction , url =

    Wang, Peng and Liu, Lingjie and Liu, Yuan and Theobalt, Christian and Komura, Taku and Wang, Wenping , booktitle =. NeuS: Learning Neural Implicit Surfaces by Volume Rendering for Multi-view Reconstruction , url =

    work page

  43. [43]

    , title =

    Hardy, Rolland L. , title =. Journal of Geophysical Research. 76 (8): 1905–1915 , year =

    work page 1905

  44. [44]

    Mathematics of Computation , year=

    Scattered data interpolation: tests of some methods , author=. Mathematics of Computation , year=

    work page

  45. [45]

    Kingma and Jimmy Ba , editor =

    Diederik P. Kingma and Jimmy Ba , editor =. Adam:. Proceedings of the International Conference on Learning Representations (ICLR) , year =

    work page

  46. [46]

    Chernoff

    Kiefer, J. and Wolfowitz, J. Stochastic Estimation of the Maximum of a Regression Function. Annals of Mathematical Statistics. 1952. doi:10.1214/aoms/1177729392

    work page doi:10.1214/aoms/1177729392 1952

  47. [47]

    2021 , url=

    Gilad Baruch and Zhuoyuan Chen and Afshin Dehghan and Tal Dimry and Yuri Feigin and Peter Fu and Thomas Gebauer and Brandon Joffe and Daniel Kurz and Arik Schwartz and Elad Shulman , booktitle=. 2021 , url=

    work page 2021

  48. [48]

    Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV) , year=

    SGAligner : 3D Scene Alignment with Scene Graphs , author=. Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV) , year=

    work page

  49. [49]

    Qianxu Wang and Haotong Zhang and Congyue Deng and Yang You and Hao Dong and Yixin Zhu and Leonidas Guibas , booktitle=. Sparse. 2024 , url=

    work page 2024

  50. [50]

    2024 , url =

    Xie, Yaxu and Pagani, Alain and Stricker, Didier , booktitle =. 2024 , url =

    work page 2024

  51. [51]

    Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , month=

    Open3DSG: Open-Vocabulary 3D Scene Graphs from Point Clouds with Queryable Objects and Open-Set Relationships , author=. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , month=

    work page

  52. [52]

    Proceedings of the International Conference on Learning Representations (ICLR) , year=

    Vision Transformers Need Registers , author=. Proceedings of the International Conference on Learning Representations (ICLR) , year=

    work page

  53. [53]

    Transactions on Machine Learning Research (TMLR) , year=

    DINOv2: Learning Robust Visual Features without Supervision , author=. Transactions on Machine Learning Research (TMLR) , year=

    work page

  54. [54]

    Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV) , year=

    Emerging Properties in Self-Supervised Vision Transformers , author=. Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV) , year=

    work page

  55. [55]

    Proceedings of the Conference on Neural Information Processing Systems (NeurIPS) , year=

    Emergent Correspondence from Image Diffusion , author=. Proceedings of the Conference on Neural Information Processing Systems (NeurIPS) , year=

    work page

  56. [56]

    Proceedings of the International Conference on Machine Learning (ICML) , year =

    Learning Transferable Visual Models From Natural Language Supervision , author =. Proceedings of the International Conference on Machine Learning (ICML) , year =

    work page

  57. [57]

    2022 , booktitle =

    CLIP-Fields: Weakly Supervised Semantic Fields for Robotic Memory , author =. 2022 , booktitle =

    work page 2022

  58. [58]

    Proceedings of the Conference on Robot Learning (CoRL) , year=

    Distilled Feature Fields Enable Few-Shot Language-Guided Manipulation , author=. Proceedings of the Conference on Robot Learning (CoRL) , year=

    work page

  59. [59]

    Proceedings of the Conference on Robot Learning (CoRL) , year=

    D ^3 Fields: Dynamic 3D Descriptor Fields for Zero-Shot Generalizable Rearrangement , author=. Proceedings of the Conference on Robot Learning (CoRL) , year=

    work page

  60. [60]

    Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) , year=

    Learning 3D Semantic Scene Graphs from 3D Indoor Reconstructions , author=. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) , year=

    work page

  61. [61]

    Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV) , year =

    3D Scene Graph: A Structure for Unified Semantics, 3D Space, and Camera , author =. Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV) , year =

    work page

  62. [62]

    Naval Research Logistics (NRL) , year=

    The Hungarian method for the assignment problem , author=. Naval Research Logistics (NRL) , year=

    work page

  63. [63]

    Murty , journal =

    Katta G. Murty , journal =. An Algorithm for Ranking all the Assignments in Order of Increasing Cost , year =

    work page

  64. [64]

    Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV) , year=

    Vector neurons: A general framework for so (3)-equivariant networks , author=. Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV) , year=

    work page

  65. [65]

    ICCV , year=

    Hyperpixel Flow: Semantic Correspondence with Multi-layer Neural Features , author=. ICCV , year=

    work page

  66. [66]

    ArXiv , year=

    SPair-71k: A Large-scale Benchmark for Semantic Correspondence , author=. ArXiv , year=

    work page

  67. [67]

    2020 , isbn =

    Jeon, Sangryul and Min, Dongbo and Kim, Seungryong and Choe, Jihwan and Sohn, Kwanghoon , title =. 2020 , isbn =. doi:10.1007/978-3-030-58604-1_38 , keywords =

    work page doi:10.1007/978-3-030-58604-1_38 2020

  68. [68]

    Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) , year=

    Telling Left from Right: Identifying Geometry-Aware Semantic Correspondence , author=. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) , year=

    work page

  69. [69]

    Proceedings of the Conference on Robot Learning (CoRL) , year=

    Automated Creation of Digital Cousins for Robust Policy Learning , author=. Proceedings of the Conference on Robot Learning (CoRL) , year=

    work page

  70. [70]

    2024 , booktitle =

    Seo, HyunA and Yi, Juheon and Balan, Rajesh and Lee, Youngki , title =. 2024 , booktitle =

    work page 2024

  71. [71]

    Sentence-BERT: Sentence Embeddings using Siamese BERT-Networks

    Reimers, Nils and Gurevych, Iryna. Sentence-BERT: Sentence Embeddings using Siamese BERT-Networks. Proceedings of the Conference on Empirical Methods in Natural Language Processing (EMNLP). 2019

    work page 2019

  72. [72]

    Moondream: A Tiny Vision Model , howpublished =. 2024

    work page 2024

  73. [73]

    Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages=

    Scaling up dynamic human-scene interaction modeling , author=. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages=

    work page

  74. [74]

    , journal=

    Umeyama, S. , journal=. Least-squares estimation of transformation parameters between two point patterns , year=

    work page

  75. [75]

    In: 2023 IEEE International Conference on Robotics and Automation (ICRA), pp

    Learning-based Relational Object Matching Across Views , author =. Proceedings of the IEEE International Conference on Robotics and Automation (ICRA) , year =. doi:10.1109/ICRA48891.2023.10161393 , url =

    work page doi:10.1109/icra48891.2023.10161393 2023

  76. [76]

    Proceedings of the Conference on Robot Learning (CoRL) , year=

    Open-TeleVision: Teleoperation with Immersive Active Visual Feedback , author=. Proceedings of the Conference on Robot Learning (CoRL) , year=

    work page

  77. [77]

    and Peluse, Patrick and Caldas, Luisa , booktitle=

    Keshavarzi, Mohammad and Zollhoefer, Michael and Yang, Allen Y. and Peluse, Patrick and Caldas, Luisa , booktitle=. Synthesizing Novel Spaces for Remote Telepresence Experiences , year=

    work page

  78. [78]

    2022 , volume=

    You, Yang and Lou, Yujing and Shi, Ruoxi and Liu, Qi and Tai, Yu-Wing and Ma, Lizhuang and Wang, Weiming and Lu, Cewu , journal=. 2022 , volume=. doi:10.1109/TPAMI.2021.3130590 , url =

    work page doi:10.1109/tpami.2021.3130590 2022

  79. [79]

    Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV) , year =

    ASIC: Aligning Sparse Image Collections , author =. Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV) , year =

    work page

  80. [80]

    arXiv preprint arXiv:2403.19474 , year=

    SG-PGM: Partial Graph Matching Network with Semantic Geometric Fusion for 3D Scene Graph Alignment and Its Downstream Tasks , author=. arXiv preprint arXiv:2403.19474 , year=

    work page arXiv

Showing first 80 references.