Pith. sign in

REVIEW 2 major objections 6 minor 1 cited by

Segment Any 3D-Part in a Scene from a Sentence

T0 review · 2 major / 6 minor · reviewed 2026-08-06 · deepseek-v4-flash

Pith's one-line read Open-vocabulary 3D part segmentation works from 3D input alone when a 2D segmentor is fine-tuned on synthetic part-annotated scenes.

desk verdict A genuinely useful synthetic part-annotated scene dataset and a solid 3D-only baseline, but the cross-dataset generalization claim is thinner than the paper's language suggests. read the letter →

arxiv 2506.19331 v1 pith:42JVJQLS submitted 2025-06-24 cs.CV

classification cs.CV
keywords open-vocabulary3Dpartsegmentationsceneunderstandingpart-levelannotationsyntheticdatasetpointcloudvision-languagemodelssuperpointgroupingtextqueries
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

Open-vocabulary 3D part segmentation asks a system to find every small part of a 3D scene that matches a free-form sentence, such as “open the washer” or “the chair legs”. This paper argues the task is feasible with 3D-only input if the 2D segmentor has been fine-tuned on a large set of synthetic scenes carrying dense part labels. To provide those labels, it introduces 3D-PU, a dataset of 10,000 scenes assembled from part-annotated 3D shapes, containing 843,654 part annotations across 206 part classes. Its OpenPart3D pipeline renders 25 views from the point cloud, segments 2D part masks with a fine-tuned vision-language model, and fuses the masks into 3D parts through weighted voting over geometric superpoints. If correct, the paper shows that costly manual part annotation in real rooms can be replaced by synthetic data, and that fine-grained part queries work on real scans without aligned RGB-D images.

What carries the argument

The mechanism that carries the argument is the view-weighted fusion of 2D part masks into 3D superpoints, fed by the Room-Tour Snap view generator. The scene is divided into a 3x3 grid; cameras are placed above each cell center and aimed at the cell centroid, with additional corner cameras, yielding 25 views designed to keep small objects and parts clearly visible. A fine-tuned 2D open-vocabulary segmentor produces per-view part masks for the text query, and each superpoint (a geometrically homogeneous point region) receives the weighted visibility score $s_i = \frac{\sum_v \sum_{p \in \hat P_i} [\mathrm{VIS}_v(p)] [\mathrm{INS}(p)] W_v}{\sum_v \sum_{p \in \hat P_i} [\mathrm{VIS}_v(p)] W_v}$, where the view weight $W_v \in \{1,2,3\}$ is higher when the camera sits in the same grid cell as the detected mask. Superpoints whose score exceeds 0.5 are grouped into the final 3D part masks, so the 2D segmentor's language understanding and the view-dependent weighting together decide what counts as a part.

What would settle it

Measure AP50 of OpenPart3D on a real dataset with dense part labels across many categories after training only on 3D-PU, and compare it with a variant whose 2D segmentor is fine-tuned on real annotated views; if synthetic fine-tuning gives no improvement or negative transfer, the central data claim fails. A second check: if 25 random camera views match the Room-Tour Snap views in AP50 on 3D-PU, then the designed camera placement is not the load-bearing mechanism.

Watch

Extended reading notes

Core claim

The paper's central claim is that part-level, open-vocabulary 3D scene understanding can be achieved by a 3D-input-only pipeline whose 2D component is fine-tuned on synthetic scenes with dense part annotations, and that this combination outperforms object-level and image-aligned alternatives on both synthetic and real benchmarks. Concretely, OpenPart3D reports AP50 of 17.8 on the 3D-PU test set and 13.7 on MultiScan, exceeding the constructed Search3D baseline at 10.7 and 10.5 respectively; on SceneFun3D functionality segmentation it reaches 9.3 AP50 without aligned 2D input, above OpenMask3D-F's 8.0 with aligned images. The paper also claims the synthetic supervision transfers: fine-tuning only the vision decoder of the 2D segmentor on rendered 3D-PU views raises AP50 by 7.4 points on 3D-PU and by 2.2 points on MultiScan, while freezing the rest of the vision-language model preserves generalization.

Load-bearing premise

The load-bearing premise is that a 2D open-vocabulary segmentor fine-tuned only on rendered views of synthetic 3D-PU scenes continues to produce accurate part masks on real-world scans, because every downstream 3D grouping decision is made from those 2D masks.

Editorial extensions

If this is right

  • Part queries can be answered from a point cloud or mesh alone, so robots and AR systems no longer need perfectly aligned RGB-D captures to find parts like handles, legs, or lids.
  • Synthetic scenes assembled from existing part-annotated shapes can act as training data for 2D open-vocabulary segmentors, lowering the cost of part-level supervision by orders of magnitude.
  • 3D-PU provides a large benchmark for open-vocabulary part segmentation and, as the paper suggests, a diagnostic dataset for 3D foundation models, analogous to CLEVR for compositional reasoning.
  • Freezing all vision-language components except the vision decoder keeps the model's open-vocabulary abilities while adapting to the synthetic domain, a recipe that should transfer to other fine-grained 3D tasks.

Reading between the lines

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

  • The paper does not test whether the same recipe works for object categories absent from its source shape collections; a natural extension is to render part-pseudo-labeled assets from any CAD collection and fine-tune only the 2D decoder, which would test whether part understanding is learned generically or memorized per category.
  • The large gap between direct queries and implicit functional queries (21.1 vs 14.3 AP50) suggests that affordance-style queries such as “open the washer” form a harder regime; a follow-up could add functional-query supervision or a separate grounding head for verbs.
  • Because view weights are assigned by hand-defined grid-cell proximity, an adaptive view planner that explicitly maximizes the expected visibility of small parts might replace the fixed 3x3 grid and improve recall on heavily occluded parts.
  • The strongest check on the synthetic-to-real claim would be an evaluation on real scenes with dense, multi-category part annotations beyond the five categories used here; the paper's current evidence for transfer rests on a narrow set of part types.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

2 major / 6 minor

Summary. The paper introduces 3D-PU, a synthetic dataset of 10,000 3D scenes with dense part-level annotations assembled from PartNet, SAPIEN, and ShapeNetPart, and proposes OpenPart3D, a pipeline that renders multiple views from a 3D mesh, obtains 2D part masks with a fine-tuned Florence2 model, and merges superpoints by weighted voting across views. The authors report AP50 of 17.8 on 3D-PU and 13.7 on MultiScan, outperforming constructed baselines, and they provide qualitative generalization results on several other datasets. The central claims are that 3D-PU is a useful large-scale resource for part-level scene understanding and that fine-tuning on it improves performance on real-world scans.

Significance. If the claims hold, 3D-PU is a potentially valuable resource for a still under-served problem, and the 3D-only input setting of OpenPart3D is practically attractive. The paper's strengths include a clearly described dataset-construction pipeline, ablations of the main components, comparisons to several baselines built from author-provided code, and qualitative demonstrations across many common 3D datasets. The dataset is assembled from existing part-annotated shape collections rather than newly annotated scenes, which limits but does not invalidate its contribution. The main weakness is that the quantitative evidence for the real-world utility of the dataset is thin: the key transfer result in Table 1 is a single comparison without variance estimates, per-category detail, or a ceiling experiment with real-data fine-tuning.

major comments (2)
  1. [Section 6 and Table 1] The claim that 3D-PU has strong practical utility for real-world data rests on the MultiScan columns of Table 1. The evidence is a single pair of point estimates: fine-tuning on 9,000 synthetic scenes improves MultiScan AP50 from 11.5 to 13.7, while the same fine-tuning improves 3D-PU AP50 by 7.4 points. There are no error bars, repeated runs, or significance tests, and no per-category breakdown for the five MultiScan part classes. I recommend adding variance estimates, a per-category table, and an upper-bound experiment in which the same vision decoder is fine-tuned on MultiScan's own 5,129 annotations. Without these, the Section 6 conclusion that fine-tuning yields 'substantial performance improvements on real-world data' is not quantitatively supported.
  2. [Section 5.1 and Table 3] The quantitative 'open-vocabulary' evaluation on MultiScan uses 47 object_part labels formed from the same 5 part classes and 17 object categories that overlap with the 206 part classes used for fine-tuning on 3D-PU. Table 3 therefore measures domain transfer on seen part classes, not the ability to segment arbitrary or unseen parts promised by the title and the task definition. Please add a quantitative evaluation on part classes or affordance labels that are absent from 3D-PU (for example, a held-out split of 3D-PU part classes, or per-affordance results on SceneFun3D), or explicitly temper the open-vocabulary claim to the taxonomy used in fine-tuning.
minor comments (6)
  1. [Figure 1 caption] The caption contains a typo: 'in s scene' should be 'in a scene'.
  2. [Table 4 caption] The caption reads 'Languge-guided functionality segmentation' and should be 'Language-guided functionality segmentation'.
  3. [Section 3] The dataset construction text uses 'GPT-4' in one place and 'ChatGPT-4' in another; please make the naming consistent.
  4. [Equation (2)] In Equation (2), the notation INS(p) does not explicitly depend on the view index v, although the following sentence describes the projection in view v. Please write INS_v(p) for clarity.
  5. [Table 4] For the SceneFun3D comparison, please clarify whether LERF and OpenMask3D-F are used with their default weights, fine-tuned on 3D-PU, or fine-tuned on SceneFun3D, since the protocol affects the interpretation of the reported gains.
  6. [Section 4] The term '3D-input-only' may be confusing because the pipeline internally renders 2D views and applies a 2D vision-language model. Please clarify that the method does not require externally captured or aligned 2D images, rather than implying that no 2D processing occurs.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the paper's central transfer claims are validated on external benchmarks not used for training.

full rationale

The paper's core contributions are empirical: construction of the 3D-PU dataset from existing part-annotated shape collections and the OpenPart3D pipeline. The fine-tuning step (Section 4, Eq. 1) trains Florence2's vision decoder on 2D masks projected from 3D-PU synthetic scenes, and the resulting model is evaluated on the 3D-PU test split, the real-world MultiScan dataset (Tables 1 and 3), and the SceneFun3D benchmark (Table 4). MultiScan and SceneFun3D contain annotations that are not used during training, so the reported improvements (e.g., Table 1: MultiScan AP50 rising from 11.5 to 13.7 after fine-tuning) are genuine transfer measurements rather than values forced by construction. No parameter in the method is fitted to the evaluation labels of MultiScan or SceneFun3D. The only taxonomic coupling is that the text queries are generated from the same object-part naming convention used for the ground-truth labels; this is a standard benchmark definition and does not by itself reduce the segmentation task, because the model must still output spatially correct masks for unseen scenes. The paper also candidly acknowledges limitations in Section 6, such as category diversity being constrained by the source datasets and layout generation requiring human validation, which are correctness considerations rather than circularity. References are to external prior work such as Florence2, SAM, PartNet, and MultiScan, and there is no load-bearing self-citation chain. Therefore, under the requirement to exhibit a specific reduction of a claimed result to its own inputs, no circular step is present.

Assumptions & free parameters 2 free parameters · 4 assumptions · 0 invented entities

The central method depends on two hand-set parameters (threshold 0.5 and view weights) and on domain assumptions about the transferability of a frozen 2D segmentor and the realism of rendered views. There are no newly invented physical or computational entities; the dataset is a collection of existing shapes with inherited annotations.

free parameters (2)
  • Superpoint foreground threshold = 0.5
    Superpoints with score s_i > 0.5 are treated as foreground (Section 4, 3D-part grouping). The threshold is hand-set and no sensitivity analysis is reported.
  • View weights W_v = 1, 2, or 3 based on grid-cell adjacency
    Views closer to the part mask receive higher weight (Section 4, View-Weighted 3D-Part Grouping). The weighting scheme is hand-designed and not optimized.
assumptions (4)
  • domain assumption The pretrained 2D open-vocabulary segmentor Florence2 can produce part-level masks for both synthetic and real images when its vision decoder is fine-tuned on synthetic renderings.
    The entire grouping step depends on the quality of 2D masks from Florence2 (Section 4, 2D part segmentation).
  • domain assumption Rendered views from the scene mesh are sufficiently similar to real RGB-D observations for the fine-tuned 2D model to transfer to real scans.
    The method is 3D-input-only and renders views; transfer to MultiScan, SceneFun3D, and other real datasets assumes the realism of textures and rendering (Sections 4 and 5).
  • domain assumption Part annotations inherited from PartNet, SAPIEN, and ShapeNetPart remain valid when shapes are textured, transformed, and placed in scenes.
    3D-PU ground truth is transferred from object-level part labels (Section 3).
  • standard math Superpoint over-segmentation aligns with part boundaries.
    Superpoints from prior work are used as the atomic units for grouping (Section 4, Superpoint generation).

how reviews work

0 comments
Cite this review

Pith. "Pith review of Segment Any 3D-Part in a Scene from a Sentence." pith.science (2026). https://pith.science/paper/42JVJQLS

@misc{pith2026250619331,
  author       = {Pith},
  title        = {Pith review of: Segment Any 3D-Part in a Scene from a Sentence},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/42JVJQLS}},
  note         = {Machine review of arXiv:2506.19331}
}
read the original abstract

This paper aims to achieve the segmentation of any 3D part in a scene based on natural language descriptions, extending beyond traditional object-level 3D scene understanding and addressing both data and methodological challenges. Due to the expensive acquisition and annotation burden, existing datasets and methods are predominantly limited to object-level comprehension. To overcome the limitations of data and annotation availability, we introduce the 3D-PU dataset, the first large-scale 3D dataset with dense part annotations, created through an innovative and cost-effective method for constructing synthetic 3D scenes with fine-grained part-level annotations, paving the way for advanced 3D-part scene understanding. On the methodological side, we propose OpenPart3D, a 3D-input-only framework to effectively tackle the challenges of part-level segmentation. Extensive experiments demonstrate the superiority of our approach in open-vocabulary 3D scene understanding tasks at the part level, with strong generalization capabilities across various 3D scene datasets.

Figures

Figures reproduced from arXiv: 2506.19331 by the authors.

Figure 1
Figure 1. The 3D-part scene understanding dataset and method. We propose the first large-scale 3D scene dataset with dense part annotations, based on an innovative approach for constructing 3D scenes with detailed part annotations (left). Using this new dataset, we introduce the 3D-part understanding task and method that enables flexible part segmentation and identification in s scene based on any sentence query (right). Abst… view at source ↗
Figure 2
Figure 2. 3D-PU dataset construction consists of the following core steps: a) collecting a large volume of 3D shapes with part annotations from PartNet [17], SAPIEN [30], and ShapeNetPart [3], b) uniformly upsampling the point clouds from meshes [34], c) generating textures and colors for each 3D shape using both an offline text-guided tex￾turing model [22] and an online texturing service, d) generating diverse layouts with t… view at source ↗
Figure 3
Figure 3. 3D-PU dataset scene examples. Each 3D scene includes a point cloud, mesh, textured mesh, and detailed part annotations for all objects. 3D Scene View-Weighted 3D-Part Grouping Module Superpoints Room-Tour Snap Module 2D Part Masks Output 2D Segmentor ... Query:“Open the washer ” Open Vocabulary 3D Part Segmentation [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗
Figures from the paper (4 more)
Figure 4
Figure 4. Figure 4: Overview of OpenPart3D. First, the Room-Tour Snap Module captures multiple view images of the 3D scene mesh by strategically positioning cameras at predefined locations with optimized poses. These images are then processed by a 2D open-vocabulary model to generate 2D p…
Figure 5
Figure 5. Figure 5: Camera position and pose. The orange circle ⃝ represents a camera positioned above the center of a grid cell, oriented toward the cell’s cen￾troid, while the blue triangle ▽ indicates a camera oriented toward the farthest corner of the scene. Within the 3×3 grid of the…
Figure 6
Figure 6. Figure 6: Qualitative examples from 3D-PU dataset. In the first scene, the small handle (in green box) of the refrigerator is accurately seg￾mented in response to the query “open the refriger￾ator.” In the second scene, most chair legs are iden￾tified, with only a few legs overl…
Figure 7
Figure 7. Figure 7: Generalization to other datasets. Our approach can be applied to a wide range of common 3D datasets, including Replica, ScanNet++, MultiScan, ARKitScenes, Matterport3D, and ScanNet. We believe it is the first time 3D part segmentation works on these datasets. 5.3 Compa…

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 1 Pith paper

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. ThinkAfford: Affordance-Centric Reasoning for Fine-Grained 3D Grounding in Cluttered Scenes

    cs.CV 2026-08 conditional novelty 6.0 of 10

    ThinkAfford decomposes 3D affordance grounding into high-recall proposal generation and GRPO-trained vision-language selection, reporting state-of-the-art AP25 on SceneFun3D.

Reference graph

Works this paper leans on

37 extracted references · 34 canonical work pages · cited by 1 Pith paper

  1. [1]

    Arkitscenes - a diverse real-world dataset for 3d indoor scene understanding using mobile RGB-d data

    Gilad Baruch, Zhuoyuan Chen, Afshin Dehghan, Tal Dimry, Yuri Feigin, Peter Fu, Thomas Gebauer, Brandon Joffe, Daniel Kurz, Arik Schwartz, and Elad Shulman. Arkitscenes - a diverse real-world dataset for 3d indoor scene understanding using mobile RGB-d data. In NeurIPS, 2021

  2. [2]

    Matterport3d: Learning from rgb-d data in indoor environments

    Angel Chang, Angela Dai, Thomas Funkhouser, Maciej Halber, Matthias Niessner, Manolis Savva, Shuran Song, Andy Zeng, and Yinda Zhang. Matterport3d: Learning from rgb-d data in indoor environments. 3DV, 2017

  3. [3]

    Shapenet: An information-rich 3d model repository

    Angel X Chang, Thomas Funkhouser, Leonidas Guibas, Pat Hanrahan, Qixing Huang, Zimo Li, Silvio Savarese, Manolis Savva, Shuran Song, Hao Su, et al. Shapenet: An information-rich 3d model repository. arXiv, 2015

  4. [4]

    Clip2scene: Towards label-efficient 3d scene understanding by clip

    Runnan Chen, Youquan Liu, Lingdong Kong, Xinge Zhu, Yuexin Ma, Yikang Li, Yuenan Hou, Yu Qiao, and Wenping Wang. Clip2scene: Towards label-efficient 3d scene understanding by clip. In CVPR, 2023

  5. [5]

    Scannet: Richly-annotated 3d reconstructions of indoor scenes

    Angela Dai, Angel X Chang, Manolis Savva, Maciej Halber, Thomas Funkhouser, and Matthias Nießner. Scannet: Richly-annotated 3d reconstructions of indoor scenes. In CVPR, 2017

  6. [6]

    SceneFun3D: Fine-grained functionality and affordance understanding in 3d scenes

    Alexandros Delitzas, Ayca Takmaz, Federico Tombari, Robert Sumner, Marc Pollefeys, and Francis Engelmann. SceneFun3D: Fine-grained functionality and affordance understanding in 3d scenes. In CVPR, 2024

  7. [7]

    Pla: Language-driven open-vocabulary 3d scene understanding

    Runyu Ding, Jihan Yang, Chuhui Xue, Wenqing Zhang, Song Bai, and Xiaojuan Qi. Pla: Language-driven open-vocabulary 3d scene understanding. In CVPR, 2023

  8. [8]

    Lowis3d: Language- driven open-world instance-level 3d scene understanding

    Runyu Ding, Jihan Yang, Chuhui Xue, Wenqing Zhang, Song Bai, and Xiaojuan Qi. Lowis3d: Language- driven open-world instance-level 3d scene understanding. TPAMI, 2024

Show all 37 references
  1. [9]

    Openins3d: Snap and lookup for 3d open-vocabulary instance segmentation

    Zhening Huang, Xiaoyang Wu, Xi Chen, Hengshuang Zhao, Lei Zhu, and Joan Lasenby. Openins3d: Snap and lookup for 3d open-vocabulary instance segmentation. In ECCV, 2024

  2. [10]

    Clevr: A diagnostic dataset for compositional language and elementary visual reasoning

    Justin Johnson, Bharath Hariharan, Laurens Van Der Maaten, Li Fei-Fei, C Lawrence Zitnick, and Ross Girshick. Clevr: A diagnostic dataset for compositional language and elementary visual reasoning. In CVPR, 2017

  3. [11]

    Lerf: Language embedded radiance fields

    Justin Kerr, Chung Min Kim, Ken Goldberg, Angjoo Kanazawa, and Matthew Tancik. Lerf: Language embedded radiance fields. In ICCV, 2023

  4. [12]

    Segment anything

    Alexander Kirillov, Eric Mintun, Nikhila Ravi, Hanzi Mao, Chloe Rolland, Laura Gustafson, Tete Xiao, Spencer Whitehead, Alexander C Berg, Wan-Yen Lo, et al. Segment anything. In ICCV, 2023

  5. [13]

    Cut pursuit: Fast algorithms to learn piecewise constant functions on general weighted graphs

    Loic Landrieu and Guillaume Obozinski. Cut pursuit: Fast algorithms to learn piecewise constant functions on general weighted graphs. Journal on Imaging Sciences, 2017

  6. [14]

    Large-scale point cloud semantic segmentation with superpoint graphs

    Loic Landrieu and Martin Simonovsky. Large-scale point cloud semantic segmentation with superpoint graphs. In CVPR, 2018

  7. [15]

    Grounding dino: Marrying dino with grounded pre-training for open-set object detection

    Shilong Liu, Zhaoyang Zeng, Tianhe Ren, Feng Li, Hao Zhang, Jie Yang, Chunyuan Li, Jianwei Yang, Hang Su, Jun Zhu, et al. Grounding dino: Marrying dino with grounded pre-training for open-set object detection. arXiv, 2023

  8. [16]

    Multiscan: Scalable rgbd scanning for 3d environments with articulated objects

    Yongsen Mao, Yiming Zhang, Hanxiao Jiang, Angel Chang, and Manolis Savva. Multiscan: Scalable rgbd scanning for 3d environments with articulated objects. In NeurIPS, 2022

  9. [17]

    Partnet: A large-scale benchmark for fine-grained and hierarchical part-level 3d object understanding

    Kaichun Mo, Shilin Zhu, Angel X Chang, Li Yi, Subarna Tripathi, Leonidas J Guibas, and Hao Su. Partnet: A large-scale benchmark for fine-grained and hierarchical part-level 3d object understanding. In CVPR, 2019

  10. [18]

    Open3DIS: Open-vocabulary 3d instance segmentation with 2d mask guidance

    Phuc Nguyen, Tuan Duc Ngo, Evangelos Kalogerakis, Chuang Gan, Anh Tran, Cuong Pham, and Khoi Nguyen. Open3DIS: Open-vocabulary 3d instance segmentation with 2d mask guidance. In CVPR, 2024

  11. [19]

    Openscene: 3d scene understanding with open vocabularies

    Songyou Peng, Kyle Genova, Chiyu Jiang, Andrea Tagliasacchi, Marc Pollefeys, Thomas Funkhouser, et al. Openscene: 3d scene understanding with open vocabularies. In CVPR, 2023

  12. [20]

    Learning transferable visual models from natural language supervision

    Alec Radford, Jong Wook Kim, Chris Hallacy, Aditya Ramesh, Gabriel Goh, Sandhini Agarwal, Girish Sastry, Amanda Askell, Pamela Mishkin, Jack Clark, et al. Learning transferable visual models from natural language supervision. In ICML, 2021. 10

  13. [21]

    Sam 2: Segment anything in images and videos

    Nikhila Ravi, Valentin Gabeur, Yuan-Ting Hu, Ronghang Hu, Chaitanya Ryali, Tengyu Ma, Haitham Khedr, Roman Rädle, Chloe Rolland, Laura Gustafson, Eric Mintun, Junting Pan, Kalyan Vasudev Alwala, Nicolas Carion, Chao-Yuan Wu, Ross Girshick, Piotr Dollár, and Christoph Feichtenh...

  14. [22]

    Texture: Text-guided texturing of 3d shapes

    Elad Richardson, Gal Metzer, Yuval Alaluf, Raja Giryes, and Daniel Cohen-Or. Texture: Text-guided texturing of 3d shapes. In SIGGRAPH, 2023

  15. [23]

    Language-grounded indoor 3d semantic segmentation in the wild

    David Rozenberszki, Or Litany, and Angela Dai. Language-grounded indoor 3d semantic segmentation in the wild. In ECCV, 2022

  16. [24]

    Mask3d: Mask transformer for 3d semantic instance segmentation

    Jonas Schult, Francis Engelmann, Alexander Hermans, Or Litany, Siyu Tang, and Bastian Leibe. Mask3d: Mask transformer for 3d semantic instance segmentation. In ICRA, 2023

  17. [25]

    Julian Straub, Thomas Whelan, Lingni Ma, Yufan Chen, Erik Wijmans, Simon Green, Jakob J. Engel, Raul Mur-Artal, Carl Ren, Shobhit Verma, Anton Clarkson, Mingfei Yan, Brian Budge, Yajie Yan, Xiaqing Pan, June Yon, Yuyang Zou, Kimberly Leon, Nigel Carter, Jesus Briales, Tyler Gi...

  18. [26]

    Going denser with open-vocabulary part segmentation

    Peize Sun, Shoufa Chen, Chenchen Zhu, Fanyi Xiao, Ping Luo, Saining Xie, and Zhicheng Yan. Going denser with open-vocabulary part segmentation. In ICCV, 2023

  19. [27]

    OpenMask3D: Open-vocabulary 3d instance segmentation

    Ayça Takmaz, Elisabetta Fedele, Robert W Sumner, Marc Pollefeys, Federico Tombari, and Francis Engelmann. OpenMask3D: Open-vocabulary 3d instance segmentation. In NeurIPS, 2023

  20. [28]

    Search3d: Hierarchical open-vocabulary 3d segmentation

    Ayca Takmaz, Alexandros Delitzas, Robert W Sumner, Francis Engelmann, Johanna Wald, and Federico Tombari. Search3d: Hierarchical open-vocabulary 3d segmentation. Robotics and Automation Letters , 2025

  21. [29]

    Partdistill: 3d shape part segmentation by vision-language model distillation

    Ardian Umam, Cheng-Kun Yang, Min-Hung Chen, Jen-Hui Chuang, and Yen-Yu Lin. Partdistill: 3d shape part segmentation by vision-language model distillation. In CVPR, 2024

  22. [30]

    Chang, Leonidas J

    Fanbo Xiang, Yuzhe Qin, Kaichun Mo, Yikuan Xia, Hao Zhu, Fangchen Liu, Minghua Liu, Hanxiao Jiang, Yifu Yuan, He Wang, Li Yi, Angel X. Chang, Leonidas J. Guibas, and Hao Su. Sapien: A simulated part-based interactive environment. In CVPR, 2020

  23. [31]

    Florence-2: Advancing a unified representation for a variety of vision tasks

    Bin Xiao, Haiping Wu, Weijian Xu, Xiyang Dai, Houdong Hu, Yumao Lu, Michael Zeng, Ce Liu, and Lu Yuan. Florence-2: Advancing a unified representation for a variety of vision tasks. In CVPR, 2024

  24. [32]

    Regionplc: Regional point-language contrastive learning for open-world 3d scene understanding

    Jihan Yang, Runyu Ding, Weipeng Deng, Zhe Wang, and Xiaojuan Qi. Regionplc: Regional point-language contrastive learning for open-world 3d scene understanding. In CVPR, 2024

  25. [33]

    Scannet++: A high-fidelity dataset of 3d indoor scenes

    Chandan Yeshwanth, Yueh-Cheng Liu, Matthias Nießner, and Angela Dai. Scannet++: A high-fidelity dataset of 3d indoor scenes. In ICCV, 2023

  26. [34]

    Sample elimination for generating poisson disk sample sets

    Cem Yuksel. Sample elimination for generating poisson disk sample sets. In Computer Graphics F orum, 2015

  27. [35]

    Clip2: Contrastive language-image-point pretraining from real-world point cloud data

    Yihan Zeng, Chenhan Jiang, Jiageng Mao, Jianhua Han, Chaoqiang Ye, Qingqiu Huang, Dit-Yan Yeung, Zhen Yang, Xiaodan Liang, and Hang Xu. Clip2: Contrastive language-image-point pretraining from real-world point cloud data. In CVPR, 2023

  28. [36]

    A simple framework for open-vocabulary segmentation and detection

    Hao Zhang, Feng Li, Xueyan Zou, Shilong Liu, Chunyuan Li, Jianwei Yang, and Lei Zhang. A simple framework for open-vocabulary segmentation and detection. In ICCV, 2023

  29. [37]

    Uni3d: Exploring unified 3d representation at scale

    Junsheng Zhou, Jinsheng Wang, Baorui Ma, Yu-Shen Liu, Tiejun Huang, and Xinlong Wang. Uni3d: Exploring unified 3d representation at scale. arXiv, 2023. 11

Pith tools

Reviewed August 6, 2026 · model on record in the stance chip above.