Pith. sign in

REVIEW 4 major objections 4 minor 58 references

ViewSRD: 3D Visual Grounding via Structured Multi-View Decomposition

T0 review · 4 major / 4 minor · reviewed 2026-08-06 · deepseek-v4-flash

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

Pith's one-line read The paper claims that 3D visual grounding improves substantially when complex multi-anchor referring expressions are first decomposed into independent single-anchor statements and processed with explicit cross-modal viewpoint tokens.

desk verdict ViewSRD's Nr3D gain is real and the method is clean, but the ScanRefer table contradicts the SOTA claim, so the paper needs revision before the numbers can be trusted. read the letter →

arxiv 2507.11261 v2 pith:HDTS2RMT submitted 2025-07-15 cs.CV

Ronggang Huang , Haoxin Yang , Yan Cai , Xuemiao Xu , Huaidong Zhang , Shengfeng He This is my paper · ORCID
classification cs.CV
keywords 3Dvisualgroundingmulti-viewdecompositionmulti-anchorquerycross-modalviewtokenspointcloudlargelanguagemodelsScanReferNr3D
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

The paper sets out to show that 3D visual grounding—finding the object in a 3D scene that a sentence names—fails for two structural reasons, and that both can be addressed before any fusion happens. Complex descriptions often name several anchor objects (for example, 'the nightstand to the right of the bed, closest to the wall with the framed picture'), and the same spatial phrase changes meaning when the observer moves. ViewSRD's claim is that rewriting such a query into a set of independent single-anchor statements, each paired with an explicit viewpoint token in both text and scene branches, removes most of the ambiguity. On Nr3D the method reports 69.9% overall accuracy, 5.2 points above the previous best under identical settings, with the largest gains on view-dependent and multi-anchor queries; on ScanRefer, the decomposition module alone improves three existing methods when added as preprocessing. If correct, the paper shows that query restructuring and viewpoint-aligned representation, rather than a better fusion head, are the levers that move accuracy.

What carries the argument

The load-bearing machinery is the pair (SRD, CCVT). SRD uses a large language model with prompt templates plus a sentence-matching filter to rewrite each multi-anchor query into one simplified single-anchor sentence per anchor, preserving the original sentence as an extra reference; this carries the disentanglement of target-anchor relations. CCVTs are learnable vectors, one per viewpoint, shared by the textual and scene modules; the text module reweights them by how well each sentence matches each viewpoint, and the scene module concatenates them to the object tokens before transformer layers. The two together make viewpoint information a common, learned currency across modalities.

What would settle it

Take a held-out set of Nr3D queries where the target's location depends on a relation between two anchors (for example, 'the chair between the table and the door that is to the left of the window') and compare ViewSRD's accuracy on those overlapping-relation queries against non-overlapping multi-anchor queries. If accuracy on overlapping queries does not improve over a baseline that keeps the original sentence, or if removing the SRD module yields equal or better accuracy on that subset, the central claim fails.

Watch

Extended reading notes

Core claim

The central discovery, stated in the paper's own terms, is that 3D visual grounding can be formulated as a structured multi-view decomposition process: first split the language, then align every piece of language with every viewpoint, then let a reasoning module vote. The Simple Relation Decoupling (SRD) module turns a multi-anchor query into one simplified sentence per anchor while keeping the original sentence as a reference; textual aggregation randomly mixes these sentences during training so the model learns from both raw and decomposed forms. The Multi-view Textual-Scene Interaction (Multi-TSI) module injects shared learnable Cross-modal Consistent View Tokens (CCVTs) into both pretrained text features and point-cloud scene features, so a phrase like 'to the right of the bed' is interpreted relative to the same viewpoint in both modalities. The final Textual-Scene Reasoning module fuses per-view predictions. Ablations show every component contributes, and removing CCVTs hurts most.

Load-bearing premise

The load-bearing premise is that a complex query can be fully rewritten as independent single-anchor sentences without losing the overlapping relationships among anchors; the paper states this assumption as its limitation, and if the rewrite drops a needed relationship the whole decomposition benefit disappears.

Editorial extensions

If this is right

  • Multi-anchor queries stop being a liability: after decomposition, the paper reports higher accuracy on multi-anchor examples (71.5%) than on single-anchor ones (69.5%) on Nr3D, the opposite of the usual trend.
  • View-dependent descriptions benefit more than view-independent ones, so viewpoint-aligned language and scene tokens are a direct cure for perspective-induced spatial ambiguity.
  • The SRD module is a drop-in text preprocessor: adding it to MVT, BUTD-DETR, and EDA improves ScanRefer accuracy without changing their architectures.
  • Better LLM decouplers push accuracy higher (70.7% with the strongest tested), so gains are tied to decomposition quality and not to the specific model.
  • Testing with more views than the four used in training continues to improve accuracy, while training with eight views plateaus, indicating a practical sweet spot around four viewpoints.

Reading between the lines

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

  • If decomposition really is the active ingredient, then language-conditioned tasks beyond grounding—such as instruction following in 3D scenes or referring-expression generation—could benefit from the same rewrite-then-align pattern; this is an extension the paper does not test.
  • A direct test would replace the LLM decoupler with a rule-based or learned relation splitter; if accuracy holds, the gain is from restructuring itself, not from LLM paraphrase quality, which changes how transferable the result is.
  • The plateau at eight views suggests uniform multi-view attention dilutes the strongest viewpoint; an adaptive view-selection mechanism, which the paper names as future work, would likely beat simply adding more cameras.
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

4 major / 4 minor

Summary. The paper proposes ViewSRD, a framework for 3D visual grounding that decomposes complex multi-anchor queries into simpler single-anchor sentences via an LLM-based Simple Relation Decoupling (SRD) module, and then fuses multi-view text and scene features using Cross-modal Consistent View Tokens (CCVTs) in a Multi-view Textual-Scene Interaction (Multi-TSI) module. Experiments on Nr3D, Sr3D, and ScanRefer report strong results on Nr3D and Sr3D (e.g., 69.9% overall on Nr3D, a 5.2-point gain over CoT3DRef), and the authors claim that the SRD module improves existing methods (MVT, BUTD-DETR, EDA) on ScanRefer. The paper also includes ablations on components, view counts, LLM decouplers, and loss weights, and releases code.

Significance. If the reported results are correct, ViewSRD would be a meaningful advance in multi-view 3D visual grounding, particularly for complex multi-anchor queries, and the SRD module as a model-agnostic preprocessing step would be a useful contribution. The paper is clearly motivated, provides code, and includes a reasonable set of ablations, including a t-SNE visualization of CCVTs. However, the ScanRefer results in Table 3 contain internal inconsistencies and arithmetic errors that directly affect the central claim that SRD enhances other methods and that ViewSRD is state-of-the-art across datasets. These issues must be resolved before the findings can be relied upon.

major comments (4)
  1. [Sec. 4.4, Table 3] The reported overall Acc@0.25 numbers in Table 3 are internally inconsistent with the stated Unique (19%) / Multiple (81%) split and the sub-scores. For BUTD-DETR+SRD, the weighted average of 85.0 (Unique) and 45.3 (Multiple) is approximately 52.8, not the reported 57.9; similarly, for BUTD-DETR the weighted average is approximately 51.9, not 50.4. Since the paper uses these overall numbers to claim a 14.9% relative improvement from SRD, the table must be corrected or the calculation explained.
  2. [Sec. 4.4, Table 3] The full ViewSRD model reports 45.4 overall Acc@0.25 on ScanRefer, which is 12.5 points lower than BUTD-DETR+SRD (57.9) and even lower than the plain BUTD-DETR baseline (50.4). Because SRD is a component of ViewSRD, this inversion is unexplained and contradicts the claim that SRD improves grounding and that ViewSRD is state-of-the-art on ScanRefer. The authors need to either provide a protocol explanation (e.g., different backbone, training data, or evaluation splits) or correct the numbers.
  3. [Sec. 4.5, Table 4 vs. Sec. 4.4, Table 3] The ablations on Nr3D (Table 4) show that removing SRD from ViewSRD costs only 1.3 points (68.6 vs. 69.9), whereas Table 3 claims SRD adds 7.5 points to BUTD-DETR on ScanRefer (50.4 to 57.9). Even accounting for dataset differences, this large discrepancy in the measured contribution of SRD is not discussed and raises questions about whether the ScanRefer improvement is replicable or confounded.
  4. [Abstract and Sec. 4.4] The abstract states that ViewSRD 'significantly outperforms state-of-the-art methods' on 3D visual grounding datasets without qualification, but Table 3 shows that on ScanRefer the full ViewSRD is outperformed by BUTD-DETR+SRD and, at Acc@0.25, is comparable to or worse than several baselines. The claim should be restricted to the datasets where it holds (Nr3D and Sr3D) or otherwise reconciled with Table 3.
minor comments (4)
  1. [Sec. 3.1 and References] The text uses 'DeepSeek-R1' but reference [27] is the DeepSeek-V3 technical report; please cite the appropriate R1 technical report or clarify the model version.
  2. [Appendix G] The appendix text says the best loss weights are 'λObj = 1.0, λRef = 0.5, λSent = 0.5', but the table columns are ordered λRef, λObj, λSent, making the row '1.0 0.5 0.5' correspond to λRef=1.0, λObj=0.5. Please fix this mismatch.
  3. [Figure 4 caption] There is a typo in the caption: 'Simple Queris' should be 'Simple Queries'.
  4. [Sec. 4.1 and Tables 1-6] All results appear to be from a single run with no error bars or variance reporting; given the small differences in Table 6 (e.g., 69.6 vs. 69.9), please state whether results are averaged over multiple seeds or report standard deviations.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: ViewSRD is an empirical architecture paper evaluated on standard held-out benchmarks, with no prediction that reduces by construction to its own inputs.

full rationale

The claimed contributions of ViewSRD are empirical: the SRD module and Multi-TSI module are trained with the stated losses on standard train/test splits of Nr3D, Sr3D, and ScanRefer, and the reported metrics are external benchmark accuracies. The SRD module's LLM-based query decomposition and the target classifier Clas are auxiliary components used for text preprocessing and training, not parameters fitted to the evaluation metric in a way that forces the reported numbers. The paper contains no load-bearing uniqueness theorem imported from the authors' prior work, no ansatz smuggled in via self-citation, and no renamed known result presented as derivation. The stated limitation that complex queries may not be fully decomposable is an acknowledged assumption, not a circular step. The internally inconsistent ScanRefer numbers in Table 3 are a correctness or comparability concern, not a circularity concern, because they do not show that any reported result is equivalent to its input by construction. Therefore the appropriate finding is no significant circularity.

Assumptions & free parameters 4 free parameters · 5 assumptions · 1 invented entities

The framework rests on standard deep learning assumptions (BERT, PointNet++, Transformer) plus two ad hoc design choices: the LLM-based query decomposition with prompt templates and the CCVT view tokens. The most fragile premise is the decomposition assumption, explicitly acknowledged by the authors as a limitation; the view token is an internal learnable construct with no external falsifiable handle.

free parameters (4)
  • Number of viewpoints N = 4
    Chosen by hand; Table 5 shows 4 views gives best trade-off and was tuned on Nr3D.
  • Loss weights (lambda_Ref, lambda_Obj, lambda_Sent) = 1.0, 0.5, 0.5 in main text; Appendix G lists a contradictory order
    Tuned on validation; Table 7 shows sensitivity to these weights.
  • Fusion coefficient alpha = sampled from {0, 0.1, 0.3, 0.5} in training, 0.5 in validation
    Chosen by hand for feature aggregation in Equation (1).
  • Length weight distribution for sentence matching = weights decaying from center 1, e.g., {0.4, 0.6, 0.8, 1, 0.8, 0.6, 0.4}
    Ad hoc formula for ideal sentence length in the SRD matching algorithm.
assumptions (5)
  • domain assumption BERT sentence features are sufficient for grounding
    Used as the text encoder without modification.
  • domain assumption PointNet++ object features are sufficient
    Used as the scene encoder.
  • domain assumption Rotating the point cloud around its central axis reproduces the viewpoint variations described in language
    Section 3 defines multi-view this way.
  • ad hoc to paper DeepSeek-R1 decomposes multi-anchor queries without altering their meaning
    The SRD module relies on this; the Conclusion acknowledges it may fail for overlapping relations.
  • domain assumption The evaluation protocol for augmented baselines in Table 3 matches the original papers
    Needed for fair comparison; no protocol details are given.
invented entities (1)
  • Cross-modal Consistent View Tokens (CCVTs)
    purpose: Learned tokens shared by text and scene modules to encode viewpoint information and align cross-modal features
    The paper provides only internal t-SNE visualization; there is no external prediction or measurement that could falsify the tokens' role.

how reviews work

0 comments
Cite this review

Pith. "Pith review of ViewSRD: 3D Visual Grounding via Structured Multi-View Decomposition." pith.science (2026). https://pith.science/paper/HDTS2RMT

@misc{pith2026250711261,
  author       = {Pith},
  title        = {Pith review of: ViewSRD: 3D Visual Grounding via Structured Multi-View Decomposition},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/HDTS2RMT}},
  note         = {Machine review of arXiv:2507.11261}
}
read the original abstract

3D visual grounding aims to identify and localize objects in a 3D space based on textual descriptions. However, existing methods struggle with disentangling targets from anchors in complex multi-anchor queries and resolving inconsistencies in spatial descriptions caused by perspective variations. To tackle these challenges, we propose ViewSRD, a framework that formulates 3D visual grounding as a structured multi-view decomposition process. First, the Simple Relation Decoupling (SRD) module restructures complex multi-anchor queries into a set of targeted single-anchor statements, generating a structured set of perspective-aware descriptions that clarify positional relationships. These decomposed representations serve as the foundation for the Multi-view Textual-Scene Interaction (Multi-TSI) module, which integrates textual and scene features across multiple viewpoints using shared, Cross-modal Consistent View Tokens (CCVTs) to preserve spatial correlations. Finally, a Textual-Scene Reasoning module synthesizes multi-view predictions into a unified and robust 3D visual grounding. Experiments on 3D visual grounding datasets show that ViewSRD significantly outperforms state-of-the-art methods, particularly in complex queries requiring precise spatial differentiation. Code is available at https://github.com/visualjason/ViewSRD.

Figures

Figures reproduced from arXiv: 2507.11261 by the authors.

Figure 1
Figure 1. (a) Previous 3DVG methods struggle with ambiguities [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. Overview of ViewSRD. We begin by employing the Simple Relation Decoupling (SRD) module to decompose complex multi￾anchor queries into multiple simpler single-anchor queries. Next, text and scene features are extracted separately using the text encoder and scene encoder. To explicitly incorporate scene information into the model, we fuse Cross-modal Consistent View Tokens (CCVTs) with these extracted features. The Mu… view at source ↗
Figure 3
Figure 3. Overview of the SRD Module pipeline. tence, respectively. This restructuring forms the foundation for generating a structured prompt, which is then fed into an LLM to produce a set of simplified queries. To maintain se￾mantic integrity, we employ a Sentence Matching algorithm which is described in detail in the supplementary material. that filters and retains the most relevant simplified queries, ensuring that the r… view at source ↗
Figures from the paper (2 more)
Figure 4
Figure 4. Figure 4: Visualization Results of the 3D Visual Grounding Results. For the presented 3D scenes, we utilize green, red, blue, and yellow [PITH_FULL_IMAGE:figures/full_fig_p006_4.png]
Figure 5
Figure 5. Figure 5: t-sne Visualization of CCVTs. G. Analysis of Loss Weight [PITH_FULL_IMAGE:figures/full_fig_p013_5.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

58 extracted references · 46 canonical work pages

  1. [1]

    Referit3d: Neural listeners for fine-grained 3d object identification in real-world scenes

    Panos Achlioptas, Ahmed Abdelreheem, Fei Xia, Mohamed Elhoseiny, and Leonidas Guibas. Referit3d: Neural listeners for fine-grained 3d object identification in real-world scenes. In ECCV, pages 422–440, 2020. 2, 6, 7, 8

    work page 2020

  2. [2]

    Cot3dref: Chain-of-thoughts data-efficient 3d visual grounding

    Eslam Mohamed Bakr, Mohamed Ayman, Mahmoud Ahmed, Habib Slim, and Mohamed Elhoseiny. Cot3dref: Chain-of-thoughts data-efficient 3d visual grounding. The Twelfth International Conference on Learning Representa- tions, 2024. 6, 7, 13

    work page 2024

  3. [3]

    Visual question answering from another perspective: Clevr mental rotation tests

    Christopher Beckham, Martin Weiss, Florian Golemo, Sina Honari, Derek Nowrouzezahrai, and Christopher Pal. Visual question answering from another perspective: Clevr mental rotation tests. Pattern Recognition, 136:109209, 2023. 4

    work page 2023

  4. [4]

    Assertiveness-based agent communica- tion for a personalized medicine on medical imaging diag- nosis

    Francisco Maria Calisto, Jo ˜ao Fernandes, Margarida Morais, Carlos Santiago, Jo˜ao Maria Abrantes, Nuno Nunes, and Jac- into C Nascimento. Assertiveness-based agent communica- tion for a personalized medicine on medical imaging diag- nosis. In Proceedings of the 2023 CHI conference on human factors in computing systems, pages 1–20, 2023. 1

    work page 2023

  5. [5]

    Mikasa: Multi-key-anchor & scene-aware transformer for 3d visual grounding

    Chun-Peng Chang, Shaoxiang Wang, Alain Pagani, and Di- dier Stricker. Mikasa: Multi-key-anchor & scene-aware transformer for 3d visual grounding. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 14131–14140, 2024. 1, 3, 7

    work page 2024

  6. [6]

    Scanrefer: 3d object localization in rgb-d scans using natural language

    Dave Zhenyu Chen, Angel X Chang, and Matthias Nießner. Scanrefer: 3d object localization in rgb-d scans using natural language. In ECCV, pages 202–221. Springer, 2020. 2, 6, 7

    work page 2020

  7. [7]

    Scjd: Sparse correlation and joint distillation for efficient 3d human pose estimation

    Weihong Chen, Xuemiao Xu, Haoxin Yang, Yi Xie, Peng Xiao, Cheng Xu, Huaidong Zhang, and Pheng-Ann Heng. Scjd: Sparse correlation and joint distillation for efficient 3d human pose estimation. arXiv preprint arXiv:2503.14097 ,

    arXiv

  8. [8]

    Talk2bev: Language-enhanced bird’s- eye view maps for autonomous driving

    Tushar Choudhary, Vikrant Dewangan, Shivam Chand- hok, Shubham Priyadarshan, Anushka Jain, Arun K Singh, Siddharth Srivastava, Krishna Murthy Jatavallabhula, and K Madhava Krishna. Talk2bev: Language-enhanced bird’s- eye view maps for autonomous driving. In 2024 IEEE In- ternational Conference on Robotics and Automation (ICRA), pages 16345–16352. IEEE, 2024. 1

    work page 2024

Show all 58 references
  1. [9]

    Drive as you speak: Enabling human-like interac- tion with large language models in autonomous vehicles

    Can Cui, Yunsheng Ma, Xu Cao, Wenqian Ye, and Ziran Wang. Drive as you speak: Enabling human-like interac- tion with large language models in autonomous vehicles. In Proceedings of the IEEE/CVF Winter Conference on Appli- cations of Computer Vision, pages 902–909, 2024. 1

    2024

  2. [10]

    Scannet: Richly-annotated 3d reconstructions of indoor scenes

    Angela Dai, Angel X Chang, Manolis Savva, Maciej Hal- ber, Thomas Funkhouser, and Matthias Nießner. Scannet: Richly-annotated 3d reconstructions of indoor scenes. In CVPR, pages 5828–5839, 2017. 2, 6

    2017

  3. [11]

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

    Jacob Devlin, Ming-Wei Chang, Kenton Lee, and Kristina Toutanova. Bert: Pre-training of deep bidirectional transformers for language understanding. arXiv preprint arXiv:1810.04805, 2018. 4, 12

    2018 arXiv

  4. [12]

    Scenegenie: Scene graph guided diffusion models for image synthesis

    Azade Farshad, Yousef Yeganeh, Yu Chi, Chengzhi Shen, B¨ojrn Ommer, and Nassir Navab. Scenegenie: Scene graph guided diffusion models for image synthesis. InICCV, pages 88–98, 2023. 2

    2023

  5. [13]

    Dense reinforce- ment learning for safety validation of autonomous vehicles

    Shuo Feng, Haowei Sun, Xintao Yan, Haojie Zhu, Zhengxia Zou, Shengyin Shen, and Henry X Liu. Dense reinforce- ment learning for safety validation of autonomous vehicles. Nature, 615(7953):620–627, 2023. 1

    2023

  6. [14]

    Viewinfer3d: 3d visual ground- ing based on embodied viewpoint inference

    Liang Geng and Jianqin Yin. Viewinfer3d: 3d visual ground- ing based on embodied viewpoint inference. IEEE Robotics and Automation Letters, 2024. 1

    2024

  7. [15]

    Viewrefer: Grasp the multi-view knowledge for 3d visual grounding

    Zoey Guo, Yiwen Tang, Ray Zhang, Dong Wang, Zhigang Wang, Bin Zhao, and Xuelong Li. Viewrefer: Grasp the multi-view knowledge for 3d visual grounding. In ICCV, pages 15372–15383, 2023. 1, 3, 7

    2023

  8. [16]

    Transrefer3d: Entity-and- relation aware transformer for fine-grained 3d visual ground- ing

    Dailan He, Yusheng Zhao, Junyu Luo, Tianrui Hui, Shaofei Huang, Aixi Zhang, and Si Liu. Transrefer3d: Entity-and- relation aware transformer for fine-grained 3d visual ground- ing. In ACM MM, pages 2344–2352, 2021. 7

    2021

  9. [17]

    3d-llm: In- jecting the 3d world into large language models

    Yining Hong, Haoyu Zhen, Peihao Chen, Shuhong Zheng, Yilun Du, Zhenfang Chen, and Chuang Gan. 3d-llm: In- jecting the 3d world into large language models. Advances in Neural Information Processing Systems, 36:20482–20494,

  10. [18]

    Multi- view transformer for 3d visual grounding

    Shijia Huang, Yilun Chen, Jiaya Jia, and Liwei Wang. Multi- view transformer for 3d visual grounding. In CVPR, pages 15524–15533, 2022. 2, 6, 7, 13

    2022

  11. [19]

    Structure-clip: Towards scene graph knowledge to enhance multi-modal structured repre- sentations

    Yufeng Huang, Jiji Tang, Zhuo Chen, Rongsheng Zhang, Xinfeng Zhang, Weijie Chen, Zeng Zhao, Zhou Zhao, Tangjie Lv, Zhipeng Hu, et al. Structure-clip: Towards scene graph knowledge to enhance multi-modal structured repre- sentations. In AAAI, pages 2417–2425, 2024. 1, 2

    2024

  12. [20]

    Nan-detr: noising multi-anchor makes detr better for object detection

    Zixin Huang, Xuesong Tao, and Xinyuan Liu. Nan-detr: noising multi-anchor makes detr better for object detection. Frontiers in Neurorobotics, 18:1484088, 2024. 1

    2024

  13. [21]

    Bottom up top down detection transform- ers for language grounding in images and point clouds

    Ayush Jain, Nikolaos Gkanatsios, Ishita Mediratta, and Kate- rina Fragkiadaki. Bottom up top down detection transform- ers for language grounding in images and point clouds. In European Conference on Computer Vision, pages 417–433. Springer, 2022. 7

    2022

  14. [22]

    Sita: Struc- turally imperceptible and transferable adversarial attacks for stylized image generation

    Jingdan Kang, Haoxin Yang, Yan Cai, Huaidong Zhang, Xuemiao Xu, Yong Du, and Shengfeng He. Sita: Struc- turally imperceptible and transferable adversarial attacks for stylized image generation. IEEE Transactions on Informa- tion Forensics and Security, 2025. 6

    2025

  15. [23]

    Is-ggt: Itera- tive scene graph generation with generative transformers

    Sanjoy Kundu and Sathyanarayanan N Aakur. Is-ggt: Itera- tive scene graph generation with generative transformers. In CVPR, pages 6292–6301, 2023. 2

    2023

  16. [24]

    Panogen: Text-conditioned panoramic environment generation for vision-and-language navigation

    Jialu Li and Mohit Bansal. Panogen: Text-conditioned panoramic environment generation for vision-and-language navigation. Advances in Neural Information Processing Sys- tems, 36, 2024. 1

    2024

  17. [25]

    Delving into in- visible semantics for generalized one-shot neural human ren- dering

    Yihong Lin, Xuemiao Xu, Huaidong Zhang, Cheng Xu, Wei- jie Li, Yi Xie, Jing Qin, and Shengfeng He. Delving into in- visible semantics for generalized one-shot neural human ren- dering. IEEE Transactions on Visualization and Computer Graphics, 2025. 2

    2025

  18. [26]

    Multi- modal situated reasoning in 3d scenes

    Xiongkun Linghu, Jiangyong Huang, Xuesong Niu, Xiao- jian Shawn Ma, Baoxiong Jia, and Siyuan Huang. Multi- modal situated reasoning in 3d scenes. Advances in Neural Information Processing Systems, 37:140903–140936, 2025. 2

    2025

  19. [27]

    Deepseek-v3 technical report

    Aixin Liu, Bei Feng, Bing Xue, Bingxuan Wang, Bochao Wu, Chengda Lu, Chenggang Zhao, Chengqi Deng, Chenyu Zhang, Chong Ruan, et al. Deepseek-v3 technical report. arXiv preprint arXiv:2412.19437, 2024. 2, 6, 8

    2024 arXiv

  20. [28]

    Rotation-adaptive point cloud domain generalization via intricate orientation learning

    Bangzhen Liu, Chenxi Zheng, Xuemiao Xu, Cheng Xu, Huaidong Zhang, and Shengfeng He. Rotation-adaptive point cloud domain generalization via intricate orientation learning. IEEE Transactions on Pattern Analysis and Ma- chine Intelligence, 2025. 2

    2025

  21. [29]

    Decoupled weight decay regularization

    Ilya Loshchilov and Frank Hutter. Decoupled weight decay regularization. arXiv preprint arXiv:1711.05101, 2017. 6

    2017 arXiv

  22. [30]

    Mmscan: A multi-modal 3d scene dataset with hierarchical grounded language annota- tions

    Ruiyuan Lyu, Jingli Lin, Tai Wang, Xiaohan Mao, Yilun Chen, Runsen Xu, Haifeng Huang, Chenming Zhu, Dahua Lin, and Jiangmiao Pang. Mmscan: A multi-modal 3d scene dataset with hierarchical grounded language annota- tions. Advances in Neural Information Processing Systems , 37:5...

    2025

  23. [31]

    Situa- tional awareness matters in 3d vision language reasoning

    Yunze Man, Liang-Yan Gui, and Yu-Xiong Wang. Situa- tional awareness matters in 3d vision language reasoning. In Proceedings of the IEEE/CVF Conference on Computer Vi- sion and Pattern Recognition, pages 13678–13688, 2024. 1

    2024

  24. [32]

    Textrank: Bringing order into text

    Rada Mihalcea and Paul Tarau. Textrank: Bringing order into text. In Proceedings of the 2004 conference on empiri- cal methods in natural language processing, pages 404–411,

    2004

  25. [33]

    Gaussian prompter: Link- ing 2d prompts for 3d gaussian segmentation

    Honghan Pan, Bangzhen Liu, Xuemiao Xu, Chenxi Zheng, Yongwei Nie, and Shengfeng He. Gaussian prompter: Link- ing 2d prompts for 3d gaussian segmentation. IEEE Trans- actions on Pattern Analysis and Machine Intelligence, 2025. 2

    2025

  26. [34]

    An approach to gener- ate a caption for an image collection using scene graph gen- eration

    Itthisak Phueaksri, Marc A Kastner, Yasutomo Kawanishi, Takahiro Komamizu, and Ichiro Ide. An approach to gener- ate a caption for an image collection using scene graph gen- eration. IEEE Access, 2023. 2

    2023

  27. [35]

    Pointnet++: Deep hierarchical feature learning on point sets in a metric space

    Charles Ruizhongtai Qi, Li Yi, Hao Su, and Leonidas J Guibas. Pointnet++: Deep hierarchical feature learning on point sets in a metric space. Advances in neural information processing systems, 30, 2017. 5

    2017

  28. [36]

    Languagerefer: Spatial-language model for 3d visual grounding

    Junha Roh, Karthik Desingh, Ali Farhadi, and Dieter Fox. Languagerefer: Spatial-language model for 3d visual grounding. In Conference on Robot Learning, pages 1046–

  29. [37]

    Aware visual grounding in 3d scenes

    Xiangxi Shi, Zhonghua Wu, and Stefan Lee. Aware visual grounding in 3d scenes. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages 14056–14065, 2024. 1

    2024

  30. [38]

    Multi-agent col- laboration: Harnessing the power of intelligent llm agents

    Yashar Talebirad and Amirhossein Nadiri. Multi-agent col- laboration: Harnessing the power of intelligent llm agents. arXiv preprint arXiv:2306.03314, 2023. 1

    2023 arXiv

  31. [39]

    Attention is all you need

    Ashish Vaswani, Noam Shazeer, Niki Parmar, Jakob Uszko- reit, Llion Jones, Aidan N Gomez, Łukasz Kaiser, and Illia Polosukhin. Attention is all you need. Advances in neural information processing systems, 30, 2017. 5

    2017

  32. [40]

    Openchat: Advancing open-source language models with mixed-quality data

    Guan Wang, Sijie Cheng, Xianyuan Zhan, Xiangang Li, Sen Song, and Yang Liu. Openchat: Advancing open-source language models with mixed-quality data. arXiv preprint arXiv:2309.11235, 2023. 2, 8

    2023 arXiv

  33. [41]

    Gˆ 3-lq: Marry- ing hyperbolic alignment with explicit semantic-geometric modeling for 3d visual grounding

    Yuan Wang, Yali Li, and Shengjin Wang. Gˆ 3-lq: Marry- ing hyperbolic alignment with explicit semantic-geometric modeling for 3d visual grounding. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 13917–13926, 2024. 4

    2024

  34. [42]

    Eda: Explicit text-decoupling and dense alignment for 3d visual grounding

    Yanmin Wu, Xinhua Cheng, Renrui Zhang, Zesen Cheng, and Jian Zhang. Eda: Explicit text-decoupling and dense alignment for 3d visual grounding. In CVPR, pages 19231– 19242, 2023. 1, 2, 6, 7

    2023

  35. [43]

    Multi- scale flow-based occluding effect and content separation for cartoon animations

    Cheng Xu, Wei Qu, Xuemiao Xu, and Xueting Liu. Multi- scale flow-based occluding effect and content separation for cartoon animations. IEEE Transactions on Visualization and Computer Graphics, 29(9):4001–4014, 2022. 6

    2022

  36. [44]

    Multi-attribute interactions matter for 3d visual grounding

    Can Xu, Yuehui Han, Rui Xu, Le Hui, Jin Xie, and Jian Yang. Multi-attribute interactions matter for 3d visual grounding. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 17253–17262, 2024. 1, 2

    2024

  37. [45]

    Learning with unreliability: Fast few-shot voxel radiance fields with relative geometric consistency

    Yingjie Xu, Bangzhen Liu, Hao Tang, Bailin Deng, and Shengfeng He. Learning with unreliability: Fast few-shot voxel radiance fields with relative geometric consistency. In Proceedings of the IEEE/CVF Conference on Computer Vi- sion and Pattern Recognition, pages 20342–20351, 2024. 2

    2024

  38. [46]

    An Yang, Baosong Yang, Beichen Zhang, Binyuan Hui, Bo Zheng, Bowen Yu, Chengyuan Li, Dayiheng Liu, Fei Huang, Haoran Wei, et al. Qwen2. 5 technical report. arXiv preprint arXiv:2412.15115, 2024. 2, 8

    2024 arXiv

  39. [47]

    G2face: High-fidelity reversible face anonymization via gen- erative and geometric priors.IEEE Transactions on Informa- tion Forensics and Security, 2024

    Haoxin Yang, Xuemiao Xu, Cheng Xu, Huaidong Zhang, Jing Qin, Yi Wang, Pheng-Ann Heng, and Shengfeng He. G2face: High-fidelity reversible face anonymization via gen- erative and geometric priors.IEEE Transactions on Informa- tion Forensics and Security, 2024. 6

    2024

  40. [48]

    Sat: 2d semantics assisted training for 3d visual grounding

    Zhengyuan Yang, Songyang Zhang, Liwei Wang, and Jiebo Luo. Sat: 2d semantics assisted training for 3d visual grounding. In ICCV, pages 1856–1866, 2021. 7

    2021

  41. [49]

    Appagent: Multimodal agents as smartphone users

    Zhao Yang, Jiaxuan Liu, Yucheng Han, Xin Chen, Zebiao Huang, Bin Fu, and Gang Yu. Appagent: Multimodal agents as smartphone users. arXiv preprint arXiv:2312.13771 ,

    arXiv

  42. [50]

    Visually-prompted language model for fine-grained scene graph generation in an open world

    Qifan Yu, Juncheng Li, Yu Wu, Siliang Tang, Wei Ji, and Yueting Zhuang. Visually-prompted language model for fine-grained scene graph generation in an open world. In ICCV, pages 21560–21571, 2023. 2

    2023

  43. [51]

    Visual programming for zero-shot open-vocabulary 3d visual grounding

    Zhihao Yuan, Jinke Ren, Chun-Mei Feng, Hengshuang Zhao, Shuguang Cui, and Zhen Li. Visual programming for zero-shot open-vocabulary 3d visual grounding. In Proceed- ings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 20623–20633, 2024. 1, 2

    2024

  44. [52]

    Multi3drefer: Grounding text description to multiple 3d ob- jects

    Yiming Zhang, ZeMing Gong, and Angel X Chang. Multi3drefer: Grounding text description to multiple 3d ob- jects. In Proceedings of the IEEE/CVF International Con- ference on Computer Vision, pages 15225–15236, 2023. 8

    2023

  45. [53]

    Towards clip-driven language-free 3d visual grounding via 2d-3d relational en- hancement and consistency

    Yuqi Zhang, Han Luo, and Yinjie Lei. Towards clip-driven language-free 3d visual grounding via 2d-3d relational en- hancement and consistency. InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages 13063–13072, 2024. 2

    2024

  46. [54]

    3dvg- transformer: Relation modeling for visual grounding on point clouds

    Lichen Zhao, Daigang Cai, Lu Sheng, and Dong Xu. 3dvg- transformer: Relation modeling for visual grounding on point clouds. In ICCV, pages 2928–2937, 2021. 7

    2021

  47. [55]

    Recdreamer: Consistent text-to-3d generation via uniform score distillation

    Chenxi Zheng, Yihong Lin, Bangzhen Liu, Xuemiao Xu, Yongwei Nie, and Shengfeng He. Recdreamer: Consistent text-to-3d generation via uniform score distillation. In The Thirteenth International Conference on Learning Represen- tations. 2

  48. [56]

    Learning an interpretable stylized subspace for 3d-aware animatable artforms

    Chenxi Zheng, Bangzhen Liu, Xuemiao Xu, Huaidong Zhang, and Shengfeng He. Learning an interpretable stylized subspace for 3d-aware animatable artforms. IEEE Transac- tions on Visualization and Computer Graphics, 31(2):1465– 1477, 2024. 2

    2024

  49. [57]

    Navgpt: Explicit reasoning in vision-and-language navigation with large lan- guage models

    Gengze Zhou, Yicong Hong, and Qi Wu. Navgpt: Explicit reasoning in vision-and-language navigation with large lan- guage models. In Proceedings of the AAAI Conference on Artificial Intelligence, pages 7641–7649, 2024. 1

    2024

  50. [58]

    Unifying 3d vision-language understanding via prompt- able queries

    Ziyu Zhu, Zhuofan Zhang, Xiaojian Ma, Xuesong Niu, Yixin Chen, Baoxiong Jia, Zhidong Deng, Siyuan Huang, and Qing Li. Unifying 3d vision-language understanding via prompt- able queries. In European Conference on Computer Vision, pages 188–206. Springer, 2024. 1 Appendix A. Sum...

    2024

Pith tools

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