pith. sign in

arxiv: 2606.08014 · v1 · pith:ADC3QOKMnew · submitted 2026-06-06 · 💻 cs.CV · cs.AI

GVC-Seg: Training-Free 3D Instance Segmentation via Geometric Visual Correspondence

Liang Xu , Fangjing Wang , Jinyu Yang , Feng Zheng This is my paper

Pith reviewed 2026-06-27 20:19 UTC · model grok-4.3

classification 💻 cs.CV cs.AI
keywords 3D instance segmentationpoint cloudtraining-freegeometric visual correspondenceensemble learningproposal aggregationopen-vocabulary segmentationCLIP features
0
0 comments X

The pith

GVC-Seg removes confidence bias from multi-model 3D instance segmentation by matching geometric and visual cues.

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

The paper presents a training-free approach to 3D instance segmentation on point clouds that aggregates proposals from multiple pre-trained models. Existing aggregation methods favor models with higher confidence scores, which vary due to differences in data handling and training. The method counters this by linking 3D geometric properties of proposals directly to corresponding 2D visual features, allowing fairer evaluation of proposal quality. Additional modules generate 3D proposals and extract mask-aware features using CLIP. This produces stronger results on standard benchmarks and extends to open-vocabulary semantic segmentation without retraining.

Core claim

GVC-Seg exploits the correspondence between 3D geometric cues and 2D visual cues to mitigate the confidence bias that arises when aggregating proposals from different segmentation models, thereby enabling unbiased ensemble learning across models whose confidence variations stem from data preprocessing and training strategies.

What carries the argument

Geometric Visual Correspondence (GVC) that aligns 3D geometric cues of proposals with 2D visual cues to assess quality without model-dependent bias, augmented by a 3D proposal generation module and a mask-aware CLIP feature extraction module.

If this is right

  • Proposal quality assessment becomes independent of individual model confidence scores.
  • State-of-the-art performance on multiple 3D instance segmentation benchmarks without training.
  • Direct applicability to open-vocabulary semantic segmentation tasks.
  • Ensemble methods can combine outputs from models trained with different strategies without additional calibration.

Where Pith is reading between the lines

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

  • The same cue-matching step could stabilize ensembles in other 3D vision tasks where model outputs carry different systematic biases.
  • If the correspondence holds across domains, it might allow mixing 2D foundation models with 3D models trained on smaller datasets.
  • The approach suggests a route to reduce reliance on post-hoc score normalization in multi-modal proposal fusion.

Load-bearing premise

The correspondence between 3D geometric cues and 2D visual cues can be established reliably enough to equalize proposal quality assessment across models.

What would settle it

Running the method on a set of proposals where one model's outputs are known to be systematically overconfident due to preprocessing differences and measuring whether its proposals still receive disproportionate selection weight.

Figures

Figures reproduced from arXiv: 2606.08014 by Fangjing Wang, Feng Zheng, Jinyu Yang, Liang Xu.

Figure 1
Figure 1. Figure 1: Illustration of the biased confidence score (C-score) and our proposed [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Overview of our GVC-Seg. The 3D Proposal Generation Module (Sec. [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Illustration of the proposed Mask-aware CLIP Feature Extraction [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Visualization of a pair of predicted instance masks from different segmentation backbones. At the top, we display a less-confident proposal with a [PITH_FULL_IMAGE:figures/full_fig_p007_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Qualitative results of three different methods in two different scenes. The original input data and segmentation results from ISBNet, Mask3D, and our [PITH_FULL_IMAGE:figures/full_fig_p008_5.png] view at source ↗
read the original abstract

Accurate 3D instance segmentation in point cloud data is critical for machine vision applications. Recent advancements leverage multiple pre-trained foundation models to generate 3D proposals, followed by the application of proposal aggregation methods, which significantly enhance performance. However, they often produce sub-optimal results due to inherent variations in confidence levels across different segmentation models, resulting in a bias toward the model with higher confidence. This bias is inherently model-dependent and is influenced by factors such as data preprocessing techniques and training strategies. To address this bias, we propose a novel, training-free 3D instance segmentation approach via Geometric Visual Correspondence (GVC-Seg), which exploits the correspondence between 3D geometric cues and 2D visual cues to mitigate the confidence bias. Additionally, a 3D proposal generation module and a mask-aware CLIP feature extraction module are introduced during the instance mask generation and instance semantic reasoning, respectively. In this way, GVC-Seg enhances proposal quality assessment, ensuring unbiased ensemble learning across different models. Extensive experiments demonstrate that our method achieves state-of-the-art performance on several challenging benchmarks, while also exhibiting strong potential in open-vocabulary semantic segmentation settings.

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

0 major / 2 minor

Summary. The manuscript presents GVC-Seg, a training-free 3D instance segmentation pipeline that aggregates proposals from multiple pre-trained foundation models by establishing geometric-visual correspondence between 3D geometric cues and 2D visual cues to remove confidence bias arising from differing preprocessing and training strategies. It introduces an explicit 3D proposal generation module and a mask-aware CLIP feature extraction module for instance mask generation and semantic reasoning, respectively, and reports state-of-the-art results on standard benchmarks together with open-vocabulary semantic segmentation capability.

Significance. If the correspondence mechanism produces reliable, unbiased proposal quality scores, the work supplies a practical, training-free ensemble strategy that exploits off-the-shelf foundation models without introducing new fitted parameters. Explicit reporting on standard benchmarks and the absence of additional training constitute reproducible strengths that would allow direct comparison with prior aggregation methods.

minor comments (2)
  1. The abstract states that extensive experiments demonstrate SOTA performance but does not name the specific benchmarks or report quantitative metrics; the experimental section should include these details for immediate verification.
  2. Notation for the geometric-visual correspondence function and the mask-aware CLIP extraction should be defined explicitly at first use to avoid ambiguity when readers compare the method to prior proposal-aggregation baselines.

Simulated Author's Rebuttal

0 responses · 0 unresolved

We thank the referee for their positive assessment of GVC-Seg and the recommendation for minor revision. The referee's summary correctly identifies the core contribution of using geometric-visual correspondence to mitigate confidence bias in a training-free ensemble of foundation models.

Circularity Check

0 steps flagged

No significant circularity; derivation relies on external pre-trained models

full rationale

The paper presents a training-free method that combines off-the-shelf foundation models with an external geometric-visual correspondence mechanism to re-score proposals. No equations, parameters, or predictions are fitted to the paper's own outputs or derived by re-labeling its inputs. The central claim rests on the reliability of the correspondence (an external assumption) rather than any self-referential reduction, self-citation chain, or ansatz smuggled from prior author work. Experiments use standard benchmarks without internal fitting loops.

Axiom & Free-Parameter Ledger

0 free parameters · 2 axioms · 0 invented entities

The approach rests on domain assumptions about model biases and cue correspondences; no free parameters or invented entities are identifiable from the abstract alone.

axioms (2)
  • domain assumption Variations in model confidence for 3D proposals arise primarily from data preprocessing and training strategies rather than other factors.
    Invoked to justify the need for bias mitigation via correspondence.
  • domain assumption Reliable geometric-visual correspondences can be computed to assess proposal quality without introducing new biases.
    Central premise of the GVC-Seg method.

pith-pipeline@v0.9.1-grok · 5740 in / 1217 out tokens · 27473 ms · 2026-06-27T20:19:44.777638+00:00 · methodology

discussion (0)

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

Reference graph

Works this paper leans on

71 extracted references · 1 canonical work pages · 1 internal anchor

  1. [1]

    Robust and Efficient RGB-D SLAM in Dynamic Environments,

    X. Yang, Z. Yuan, D. Zhu, C. Chi, K. Li, and C. Liao, “Robust and Efficient RGB-D SLAM in Dynamic Environments,”IEEE Transactions on Multimedia (TMM), vol. 23, pp. 4208–4219, 2020

    2020

  2. [2]

    Scene Recognition Mechanism for Service Robot Adapting Various Families: A CNN-Based Approach Using Multi-Type Cameras,

    S. Liu, G. Tian, Y . Zhang, and P. Duan, “Scene Recognition Mechanism for Service Robot Adapting Various Families: A CNN-Based Approach Using Multi-Type Cameras,”IEEE Transactions on Multimedia (TMM), vol. 24, pp. 2392–2406, 2021

    2021

  3. [3]

    Real-time 3D Single Object Tracking with Transformer,

    J. Shan, S. Zhou, Y . Cui, and Z. Fang, “Real-time 3D Single Object Tracking with Transformer,”IEEE Transactions on Multimedia (TMM), vol. 25, pp. 2339–2353, 2022

    2022

  4. [4]

    3D Object Segmentation using Cross-Window Point Transformer with Latent Semantic Boundary Guidance,

    Q. Wang, D. Liu, Z. Liu, J. Xu, and J. Tan, “3D Object Segmentation using Cross-Window Point Transformer with Latent Semantic Boundary Guidance,”IEEE Transactions on Multimedia (TMM), 2023

    2023

  5. [5]

    LIF-Seg: LiDAR and Camera Image Fusion for 3D LiDAR Semantic Segmenta- tion,

    L. Zhao, H. Zhou, X. Zhu, X. Song, H. Li, and W. Tao, “LIF-Seg: LiDAR and Camera Image Fusion for 3D LiDAR Semantic Segmenta- tion,”IEEE Transactions on Multimedia (TMM), vol. 26, pp. 1158–1168, 2023

    2023

  6. [6]

    Unsupervised Point Cloud Co-part Segmentation via Co-attended Superpoint Genera- tion and Aggregation,

    A. Umam, C.-K. Yang, J.-H. Chuang, and Y .-Y . Lin, “Unsupervised Point Cloud Co-part Segmentation via Co-attended Superpoint Genera- tion and Aggregation,”IEEE Transactions on Multimedia (TMM), 2024

    2024

  7. [7]

    PointGT: A Method for Point-Cloud Classification and Segmentation Based on Local Geometric Transformation,

    H. Zhang, C. Wang, L. Yu, S. Tian, X. Ning, and J. Rodrigues, “PointGT: A Method for Point-Cloud Classification and Segmentation Based on Local Geometric Transformation,”IEEE Transactions on Multimedia (TMM), 2024

    2024

  8. [8]

    Multi-view Convolutional Neural Networks for 3D Shape Recognition,

    H. Su, S. Maji, E. Kalogerakis, and E. Learned-Miller, “Multi-view Convolutional Neural Networks for 3D Shape Recognition,” inProc. ICCV, 2015, pp. 945–953

    2015

  9. [9]

    PAConv: Position Adaptive Convolution with Dynamic Kernel Assembling on Point Clouds,

    M. Xu, R. Ding, H. Zhao, and X. Qi, “PAConv: Position Adaptive Convolution with Dynamic Kernel Assembling on Point Clouds,” in Proc. CVPR, 2021, pp. 3173–3182

    2021

  10. [10]

    Cylindrical and Asymmetrical 3D Convolution Networks for LiDAR-based Perception,

    X. Zhu, H. Zhou, T. Wang, F. Hong, W. Li, Y . Ma, H. Li, R. Yang, and D. Lin, “Cylindrical and Asymmetrical 3D Convolution Networks for LiDAR-based Perception,”IEEE Trans. Pattern Anal. Mach. Intell., vol. 44, no. 10, pp. 6807–6822, 2021

    2021

  11. [11]

    V oxNet: A 3D Convolutional Neural Network for real-time object recognition,

    D. Maturana and S. Scherer, “V oxNet: A 3D Convolutional Neural Network for real-time object recognition,” inProc. IROS. IEEE, 2015, pp. 922–928

    2015

  12. [12]

    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,” inProc. CVPR, 2017, pp. 77–85

    2017

  13. [13]

    PointNet++: Deep Hierarchical Feature Learning on Point Sets in a Metric Space,

    C. R. Qi, L. Yi, H. Su, and L. J. Guibas, “PointNet++: Deep Hierarchical Feature Learning on Point Sets in a Metric Space,” inProc. NeurIPS, vol. 30, 2017, pp. 5099–5108

    2017

  14. [14]

    Frustum PointNets for 3D Object Detection from RGB-D Data,

    C. R. Qi, W. Liu, C. Wu, H. Su, and L. J. Guibas, “Frustum PointNets for 3D Object Detection from RGB-D Data,” inProc. CVPR, 2018, pp. 918–927

    2018

  15. [15]

    Attention Is All You Need,

    A. Vaswani, N. Shazeer, N. Parmar, J. Uszkoreit, L. Jones, A. N. Gomez, Ł. Kaiser, and I. Polosukhin, “Attention Is All You Need,”Proc. NeurIPS, vol. 30, 2017

    2017

  16. [16]

    Point Transformer,

    H. Zhao, L. Jiang, J. Jia, P. H. Torr, and V . Koltun, “Point Transformer,” inProc. ICCV, 2021, pp. 16 259–16 268

    2021

  17. [17]

    Mask3D for 3D Semantic Instance Segmentation,

    J. Schult, F. Engelmann, A. Hermans, O. Litany, S. Tang, and B. Leibe, “Mask3D for 3D Semantic Instance Segmentation,” inProc. ICRA, 2023

    2023

  18. [18]

    ISBNet: a 3D Point Cloud Instance Segmentation Network with Instance-aware Sampling and Box- aware Dynamic Convolution,

    T. D. Ngo, B.-S. Hua, and K. Nguyen, “ISBNet: a 3D Point Cloud Instance Segmentation Network with Instance-aware Sampling and Box- aware Dynamic Convolution,” inProc. CVPR, 2023, pp. 13 550–13 559

    2023

  19. [19]

    ScanNet: Richly-annotated 3D Reconstructions of Indoor Scenes,

    A. Dai, A. X. Chang, M. Savva, M. Halber, T. Funkhouser, and M. Nießner, “ScanNet: Richly-annotated 3D Reconstructions of Indoor Scenes,” inProc. CVPR, 2017

    2017

  20. [20]

    ScanNet++: A High- Fidelity Dataset of 3D Indoor Scenes,

    C. Yeshwanth, Y .-C. Liu, M. Nießner, and A. Dai, “ScanNet++: A High- Fidelity Dataset of 3D Indoor Scenes,” inProc. ICCV, 2023

    2023

  21. [21]

    Language-Grounded Indoor 3D Semantic Segmentation in the Wild,

    D. Rozenberszki, O. Litany, and A. Dai, “Language-Grounded Indoor 3D Semantic Segmentation in the Wild,” inProc. ECCV, 2022

    2022

  22. [22]

    SAMPro3D: Locating SAM Prompts in 3D for Zero-Shot Instance Segmentation,

    M. Xu, X. Yin, L. Qiu, Y . Liu, X. Tong, and X. Han, “SAMPro3D: Locating SAM Prompts in 3D for Zero-Shot Instance Segmentation,” in Proc. 3DV, 2025

    2025

  23. [23]

    SAM3D: Segment Anything in 3D Scenes,

    Y . Yang, X. Wu, T. He, H. Zhao, and X. Liu, “SAM3D: Segment Anything in 3D Scenes,” inProc. ICCVW, 2023

    2023

  24. [24]

    OVIR-3D: Open-V ocabulary 3D Instance Retrieval Without Training on 3D Data,

    S. Lu, H. Chang, E. P. Jing, A. Boularias, and K. Bekris, “OVIR-3D: Open-V ocabulary 3D Instance Retrieval Without Training on 3D Data,” inProc. CoRL, 2023

    2023

  25. [25]

    Function- ality Understanding and Segmentation in 3D Scenes,

    J. Corsetti, F. Giuliari, A. Fasoli, D. Boscaini, and F. Poiesi, “Function- ality Understanding and Segmentation in 3D Scenes,” inProc. CVPR, 2025, pp. 24 550–24 559

    2025

  26. [26]

    Open3DIS: Open-V ocabulary 3D Instance Segmentation with 2D Mask Guidance,

    P. Nguyen, T. D. Ngo, E. Kalogerakis, C. Gan, A. Tran, C. Pham, and K. Nguyen, “Open3DIS: Open-V ocabulary 3D Instance Segmentation with 2D Mask Guidance,” inProc. CVPR, 2024, pp. 4018–4028

    2024

  27. [27]

    SAI3D: Segment Any Instance in 3D Scenes,

    Y . Yin, Y . Liu, Y . Xiao, D. Cohen-Or, J. Huang, and B. Chen, “SAI3D: Segment Any Instance in 3D Scenes,” inProc. CVPR, 2024, pp. 3292– 3302

    2024

  28. [28]

    OpenMask3D: Open-V ocabulary 3D Instance Segmen- tation,

    A. Takmaz, E. Fedele, R. W. Sumner, M. Pollefeys, F. Tombari, and F. Engelmann, “OpenMask3D: Open-V ocabulary 3D Instance Segmen- tation,” inProc. NeurIPS, 2023

    2023

  29. [29]

    3D Part Segmentation via Geometric Aggregation of 2D Visual Fea- tures,

    M. Garosi, R. Tedoldi, D. Boscaini, M. Mancini, N. Sebe, and F. Poiesi, “3D Part Segmentation via Geometric Aggregation of 2D Visual Fea- tures,” inProc. WACV. IEEE, 2025, pp. 3257–3267

    2025

  30. [30]

    EmbodiedSAM: Online Segment Any 3D Thing in Real Time,

    X. Xu, H. Chen, L. Zhao, Z. Wang, J. Zhou, and J. Lu, “EmbodiedSAM: Online Segment Any 3D Thing in Real Time,” inProc. ICLR, 2025

    2025

  31. [31]

    SGS-3D: High- Fidelity 3D Instance Segmentation via Reliable Semantic Mask Splitting and Growing,

    C. Wang, Y . Luo, J. Du, S. Chen, Y . Chen, and T. Han, “SGS-3D: High- Fidelity 3D Instance Segmentation via Reliable Semantic Mask Splitting and Growing,” inProc. AAAI, 2025

    2025

  32. [32]

    Learning Transferable Visual Models From Natural Language Supervision,

    A. Radford, J. W. Kim, C. Hallacy,et al., “Learning Transferable Visual Models From Natural Language Supervision,” inProc. ICML. PMLR, 2021, pp. 8748–8763

    2021

  33. [33]

    The Replica Dataset: A Digital Replica of Indoor Spaces

    J. Straub, T. Whelan, L. Ma, Y . Chen, E. Wijmans, S. Green, J. J. Engel, R. Mur-Artal, C. Ren, S. Vermaet al., “The Replica Dataset: A Digital Replica of Indoor Spaces,”arXiv preprint arXiv:1906.05797, 2019

    work page internal anchor Pith review Pith/arXiv arXiv 1906

  34. [34]

    4D Spatio-Temporal ConvNets: Minkowski Convolutional Neural Networks,

    C. Choy, J. Gwak, and S. Savarese, “4D Spatio-Temporal ConvNets: Minkowski Convolutional Neural Networks,” inProc. CVPR, 2019, pp. 3075–3084

    2019

  35. [35]

    SCF- Net: Learning Spatial Contextual Features for Large-Scale Point Cloud Segmentation,

    S. Fan, Q. Dong, F. Zhu, Y . Lv, P. Ye, and F.-Y . Wang, “SCF- Net: Learning Spatial Contextual Features for Large-Scale Point Cloud Segmentation,” inProc. CVPR, 2021, pp. 14 504–14 513. PREPRINT SUBMITTED TO IEEE 10

    2021

  36. [36]

    Superpoint Transformer for 3D Scene Instance Segmentation,

    J. Sun, C. Qing, J. Tan, and X. Xu, “Superpoint Transformer for 3D Scene Instance Segmentation,” inProc. AAAI, vol. 37, no. 2, 2023, pp. 2393–2401

    2023

  37. [37]

    PointConv: Deep Convolutional Networks on 3D Point Clouds,

    W. Wu, Z. Qi, and L. Fuxin, “PointConv: Deep Convolutional Networks on 3D Point Clouds,” inProc. CVPR, 2019, pp. 9621–9630

    2019

  38. [38]

    Dynamic Graph CNN for Learning on Point Clouds,

    Y . Wang, Y . Sun, Z. Liu, S. E. Sarma, M. M. Bronstein, and J. M. Solomon, “Dynamic Graph CNN for Learning on Point Clouds,”ACM Transactions on Graphics (TOG), vol. 38, no. 5, pp. 1–12, 2019

    2019

  39. [39]

    PointCNN: Convolution On X-Transformed Points,

    Y . Li, R. Bu, M. Sun, W. Wu, X. Di, and B. Chen, “PointCNN: Convolution On X-Transformed Points,”Proc. NeurIPS, vol. 31, 2018

    2018

  40. [40]

    Virtual Multi-view Fusion for 3D Semantic Segmen- tation,

    A. Kundu, X. Yin, A. Fathi, D. Ross, B. Brewington, T. Funkhouser, and C. Pantofaru, “Virtual Multi-view Fusion for 3D Semantic Segmen- tation,” inProc. ECCV. Springer, 2020, pp. 518–535

    2020

  41. [41]

    Top- Down Beats Bottom-Up in 3D Instance Segmentation,

    M. Kolodiazhnyi, A. V orontsova, A. Konushin, and D. Rukhovich, “Top- Down Beats Bottom-Up in 3D Instance Segmentation,” inProc. WACV, 2024, pp. 3566–3574

    2024

  42. [42]

    3D Semantic Segmentation with Submanifold Sparse Convolutional Networks,

    B. Graham, M. Engelcke, and L. Van Der Maaten, “3D Semantic Segmentation with Submanifold Sparse Convolutional Networks,” in Proc. CVPR, 2018, pp. 9224–9232

    2018

  43. [43]

    SEGCloud: Semantic Segmentation of 3D Point Clouds,

    L. Tchapmi, C. Choy, I. Armeni, J. Gwak, and S. Savarese, “SEGCloud: Semantic Segmentation of 3D Point Clouds,” inProc. 3DV. IEEE, 2017, pp. 537–547

    2017

  44. [44]

    KPConv: Flexible and Deformable Convolution for Point Clouds,

    H. Thomas, C. R. Qi, J.-E. Deschaud, B. Marcotegui, F. Goulette, and L. J. Guibas, “KPConv: Flexible and Deformable Convolution for Point Clouds,” inProc. ICCV, 2019, pp. 6411–6420

    2019

  45. [45]

    GrowSP: Unsupervised Semantic Segmentation of 3D Point Clouds,

    Z. Zhang, B. Yang, B. Wang, and B. Li, “GrowSP: Unsupervised Semantic Segmentation of 3D Point Clouds,” inProc. CVPR, 2023, pp. 17 619–17 629

    2023

  46. [46]

    Scalable softgroup for 3d instance segmentation on point clouds,

    T. Vu, K. Kim, T. Nguyen, T. M. Luu, J. Kim, and C. D. Yoo, “Scalable softgroup for 3d instance segmentation on point clouds,”IEEE Trans. Pattern Anal. Mach. Intell., vol. 46, no. 4, pp. 1981–1995, 2024

    1981

  47. [47]

    RandLA-Net: Efficient Semantic Segmentation of Large- Scale Point Clouds,

    Q. Hu, B. Yang, L. Xie, S. Rosa, Y . Guo, Z. Wang, N. Trigoni, and A. Markham, “RandLA-Net: Efficient Semantic Segmentation of Large- Scale Point Clouds,” inProc. CVPR, 2020, pp. 11 108–11 117

    2020

  48. [48]

    Scaling up visual and vision-language representation learning with noisy text supervision,

    C. Jia, Y . Yang, Y . Xiaet al., “Scaling up visual and vision-language representation learning with noisy text supervision,” inProc. ICML, vol

  49. [49]

    4904–4916

    PMLR, 2021, pp. 4904–4916

    2021

  50. [50]

    Language-driven Semantic Segmentation,

    B. Li, K. Q. Weinberger, S. Belongie, V . Koltun, and R. Ranftl, “Language-driven Semantic Segmentation,” inProc. ICLR, 2022

    2022

  51. [51]

    Emerging Properties in Self-Supervised Vision Transform- ers,

    M. Caron, H. Touvron, I. Misra, H. J ´egou, J. Mairal, P. Bojanowski, and A. Joulin, “Emerging Properties in Self-Supervised Vision Transform- ers,” inProc. ICCV, 2021, pp. 9650–9660

    2021

  52. [52]

    VisualBERT: A Simple and Performant Baseline for Vision and Language,

    L. Li, M. Yatskar, D. Yin, C.-J. Hsieh, and K.-W. Chang, “VisualBERT: A Simple and Performant Baseline for Vision and Language,” inProc. EMNLP, 2019, pp. 3293–3303

    2019

  53. [53]

    ViLBERT: Pretraining Task-Agnostic Visiolinguistic Representations for Vision-and-Language Tasks,

    J. Lu, D. Batra, D. Parikh, and S. Lee, “ViLBERT: Pretraining Task-Agnostic Visiolinguistic Representations for Vision-and-Language Tasks,” inProc. NeurIPS, 2019, pp. 13–23

    2019

  54. [54]

    Denseclip: Language-guided dense prediction with context-aware prompting,

    Y . Rao, W. Zhao, G. Chen, Y . Tang, Z. Zhu, G. Huang, J. Zhou, and J. Lu, “Denseclip: Language-guided dense prediction with context-aware prompting,” inProc. CVPR, 2022, pp. 18 082–18 091

    2022

  55. [55]

    A Simple Baseline for Open-V ocabulary Semantic Segmentation with Pre-trained Vision-language Model,

    M. Xu, Z. Zhang, F. Wei, Y . Lin, Y . Cao, H. Hu, and X. Bai, “A Simple Baseline for Open-V ocabulary Semantic Segmentation with Pre-trained Vision-language Model,” inProc. ECCV. Springer, 2022, pp. 736–753

    2022

  56. [56]

    WinCLIP: Zero-/Few-Shot Anomaly Classification and Segmentation,

    J. Jeong, Y . Zou, T. Kim, D. Zhang, A. Ravichandran, and O. Dabeer, “WinCLIP: Zero-/Few-Shot Anomaly Classification and Segmentation,” inProc. CVPR, 2023, pp. 19 606–19 616

    2023

  57. [57]

    Detecting Everything in the Open World: Towards Universal Object Detection,

    Z. Wang, Y . Li, X. Chen, S.-N. Lim, A. Torralba, H. Zhao, and S. Wang, “Detecting Everything in the Open World: Towards Universal Object Detection,” inProc. CVPR, 2023, pp. 11 433–11 443

    2023

  58. [58]

    Simple Open-V ocabulary Object Detection,

    M. Minderer, A. Gritsenko, A. Stoneet al., “Simple Open-V ocabulary Object Detection,” inProc. ECCV. Springer, 2022, pp. 728–755

    2022

  59. [59]

    Openscene: 3d scene understanding with open vocabularies,

    S. Peng, K. Genova, C. M. Jiang, A. Tagliasacchi, M. Pollefeys, and T. Funkhouser, “Openscene: 3d scene understanding with open vocabularies,” inProc. CVPR, 2023

    2023

  60. [60]

    OpenIns3D: Snap and Lookup for 3D Open-vocabulary Instance Seg- mentation,

    Z. Huang, X. Wu, X. Chen, H. Zhao, L. Zhu, and J. Lasenby, “OpenIns3D: Snap and Lookup for 3D Open-vocabulary Instance Seg- mentation,” inProc. ECCV, 2024

    2024

  61. [61]

    Segment Anything,

    A. Kirillov, E. Mintun, N. Ravi, H. Mao, C. Rolland, L. Gustafson, T. Xiao, S. Whitehead, A. C. Berg, W.-Y . Loet al., “Segment Anything,” inProc. ICCV, 2023, pp. 4015–4026

    2023

  62. [62]

    Decoupling Zero-Shot Semantic Segmentation,

    J. Ding, N. Xue, G.-S. Xia, and D. Dai, “Decoupling Zero-Shot Semantic Segmentation,” inProc. CVPR, 2022, pp. 11 583–11 592

    2022

  63. [63]

    GroupViT: Semantic Segmentation Emerges from Text Supervision,

    J. Xu, S. De Mello, S. Liu, W. Byeon, T. Breuel, J. Kautz, and X. Wang, “GroupViT: Semantic Segmentation Emerges from Text Supervision,” in Proc. CVPR, 2022, pp. 18 134–18 144

    2022

  64. [64]

    Zero-Shot Semantic Segmentation,

    M. Bucher, T.-H. Vu, M. Cord, and P. P ´erez, “Zero-Shot Semantic Segmentation,” inProc. NeurIPS, vol. 32, 2019

    2019

  65. [65]

    Segment anything in 3D with NeRFs,

    J. Cen, Z. Zhou, J. Fang, W. Shen, L. Xie, D. Jiang, X. Zhang, Q. Tian et al., “Segment anything in 3D with NeRFs,”Proc. NeurIPS, vol. 36, pp. 25 971–25 990, 2023

    2023

  66. [66]

    When 3D Bounding-Box Meets SAM: Point Cloud Instance Segmentation with Weak-and-Noisy Supervision,

    Q. Yu, H. Du, C. Liu, and X. Yu, “When 3D Bounding-Box Meets SAM: Point Cloud Instance Segmentation with Weak-and-Noisy Supervision,” inProc. WACV, 2024, pp. 3719–3728

    2024

  67. [67]

    Grounding DINO: Marrying DINO with Grounded Pre- Training for Open-Set Object Detection,

    S. Liu, Z. Zeng, T. Ren, F. Li, H. Zhang, J. Yang, Q. Jiang, C. Li, J. Yang, H. Suet al., “Grounding DINO: Marrying DINO with Grounded Pre- Training for Open-Set Object Detection,” inProc. ECCV. Springer, 2024, pp. 38–55

    2024

  68. [68]

    Yolov9: Learning what you want to learn using programmable gradient information,

    C.-Y . Wang, I.-H. Yeh, and H.-Y . Mark Liao, “Yolov9: Learning what you want to learn using programmable gradient information,” inProc. ECCV. Springer, 2024, pp. 1–21

    2024

  69. [69]

    Microsoft COCO: Common Objects in Context,

    T.-Y . Lin, M. Maire, S. Belongie, J. Hays, P. Perona, D. Ramanan, P. Doll ´ar, and C. L. Zitnick, “Microsoft COCO: Common Objects in Context,” inProc. ECCV. Springer, 2014, pp. 740–755

    2014

  70. [70]

    A Density-Based Algorithm for Discovering Clusters in Large Spatial Databases with Noise,

    M. Ester, H.-P. Kriegel, J. Sander, X. Xuet al., “A Density-Based Algorithm for Discovering Clusters in Large Spatial Databases with Noise,” inProc. KDD, vol. 96, no. 34, 1996, pp. 226–231

    1996

  71. [71]

    MaskClustering: View Consensus Based Mask Graph Clustering for Open-V ocabulary 3D Instance Segmentation,

    M. Yan, J. Zhang, Y . Zhu, and H. Wang, “ MaskClustering: View Consensus Based Mask Graph Clustering for Open-V ocabulary 3D Instance Segmentation,” inProc. CVPR, Jun. 2024, pp. 28 274–28 284. － 乙 ． Liang Xureceived his Bachelor’s degree in Biomed- ical Engineering from Xidian University, Xi’an, China, in 2016, and his M.S. degree in Information and Commu...

    2024