TDVR: Joint Text Disambiguation and Viewpoint Reasoning for Zero-Shot 3D Visual Grounding

Junbo Wang; Qingxi Du; Yining Zhu; Yuke Li

REVIEW 4 major objections 5 minor 47 references

TDVR: Joint Text Disambiguation and Viewpoint Reasoning for Zero-Shot 3D Visual Grounding

T0 review · 4 major / 5 minor · reviewed 2026-08-05 · deepseek-v4-flash

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

Pith's one-line read This paper claims that a training-free pipeline, TDVR, sets a new state-of-the-art for zero-shot 3D visual grounding by jointly disambiguating queries and inferring the observer's horizontal viewpoint.

desk verdict TDVR is a well-built training-free pipeline for zero-shot 3D grounding with plausible module-level gains, but the headline SOTA needs to be re-verified under controlled baselines and with validation-based hyperparameter selection. read the letter →

arxiv 2608.03763 v1 pith:Z7XHHHNS submitted 2026-08-04 cs.CV

Qingxi Du , Junbo Wang , Yuke Li , Yining Zhu This is my paper

classification cs.CV

keywords zero-shot3DvisualgroundingscenegraphviewpointinferencetextdisambiguationpointcloudlargelanguagemodelspatialreasoningScanRefer

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

TDVR is a training-free framework for zero-shot 3D visual grounding that tackles two specific failure modes: ambiguous referring text and missing viewpoint information. The paper argues that both can be resolved by reasoning rather than learning: a semantic scene graph is built from detected instances, a large language model enriches and disambiguates the query, and the observer's viewpoint is reduced to a one-dimensional horizontal rotation search. The final target is scored by combining category, appearance, viewpoint, and within-category confusion scores. If the reported numbers hold, this zero-shot method outperforms all prior zero-shot baselines and several fully supervised models on ScanRefer, narrowing the gap between training-free and trained 3D grounding.

What carries the argument

Two mechanisms carry the argument. First, viewpoint-aware directional reasoning uses rotation invariance: rotating both objects around any center c transforms the relative displacement v into v' = R v, which does not depend on c; with the prior that vertical (Z-axis) relations need no pitch optimization, a full 6-D viewpoint search collapses to a 1-D yaw sweep. This sweep produces the viewpoint score S_v = max_θ Σ_j Ψ(T, C_j, θ), aligning the target-to-anchor vectors to reference directions. Second, a view-based similarity decoupled reasoning module compares each candidate's displacement from its category centroid to the intra-category direction parsed from text, yielding a confusion score S

What would settle it

Run the framework on a held-out set of ScanRefer/Sr3D queries that describe vertical relations in a viewpoint-dependent way, such as 'the chair above the table when seen from below,' while an otherwise identical variant also searches pitch over a small range; if the pitch-aware variant gives materially higher Acc@0.5, the yaw-only reduction is the limiting assumption.

Watch

Extended reading notes

Core claim

TDVR claims that zero-shot 3D visual grounding degrades for two fixable reasons—ambiguous referring text and missing viewpoint information—and that both can be addressed without any training. It constructs a semantic scene graph from detected instances, asks a large language model to enrich the original query with appearance details, anchor directions, and intra-category positions, then parses the enriched text into structured relations. The geometric core is the reduction of viewpoint inference to a 1-D horizontal rotation search: because relative object vectors transform as v' = R v independent of the rotation center, and vertical relations are taken as unambiguous, the optimal camera yaw

Load-bearing premise

The load-bearing assumption is that vertical spatial relations are unambiguous, so the observer's viewpoint is fully captured by a single horizontal rotation angle; if pitch, gaze elevation, or occlusion-constrained visibility matters, the computed viewpoint and confusion scores are systematically wrong.

Editorial extensions

If this is right

TDVR reaches 64.06% Acc@0.5 and 70.85% Acc@0.25 on ScanRefer, surpassing all previous zero-shot baselines and several fully supervised methods without any training.
On the Sr3D view-dependent subset it reaches 79.03%, showing explicit viewpoint reasoning handles directional language reliably.
Ablations show removing the observer-centric disambiguation drops accuracy by 23.4%, identifying ambiguous query text as the dominant error source.
Viewpoint-aware directional reasoning contributes 19.3% and the view-based similarity decoupled reasoning adds 3.2%, confirming that both viewpoint and fine-grained intra-class discrimination matter.
The framework processes a query in 10.21 seconds on average, faster than SPAZER (23.5s) and Vlm-Grounder (50.3s), making training-free grounding practical for interactive agents.

Reading between the lines

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

The yaw-only reduction assumes vertical relations are always resolvable by Z-axis ordering; a natural extension is a 2-D yaw+pitch search for embodied agents with head tilt, targeting descriptions like 'the shelf above the sink seen from a low viewpoint.'
Observer-centric disambiguation enriches queries with LLM-generated anchor descriptions; this enrichment could be tested as a generic pre-processing step for any 3DVG model, supervised or zero-shot, and measured for transfer across backbones.
The ablation's 23.4% drop without disambiguation suggests query-language uncertainty, not geometry, is currently the main bottleneck in zero-shot 3DVG; later systems may gain more from language-side improvements than scene-side ones.
The confusion score uses the same-category centroid as reference, which degenerates when only two instances exist (the centroid becomes the midpoint); fusing nearest-neighbor relative vectors instead of the centroid is a testable improvement.

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.

Desk Editor's Note

read the letter

Quick take: TDVR is a solid, training-free pipeline for zero-shot 3D visual grounding. The combination of observer-centric disambiguation, viewpoint-aware directional scoring, and intra-class confusion scoring is genuinely new in this subfield, and the ablations support each module's contribution. The reported gains on ScanRefer are large and the inference time is competitive.

What's good: The architecture is sensible. The scene graph construction with occlusion-aware 2D crops is careful. The disambiguation step uses an LLM to enrich the query with appearance and spatial descriptions, which makes sense. The viewpoint reasoning rotates the point cloud and computes directional alignment, and the confusion score handles same-class distractors. The ablation table shows each module matters, with disambiguation and viewpoint reasoning being the big contributors. The paper also compares to a range of recent zero-shot and supervised methods, and reports inference time. That is a useful data point.

Where I'd push back: The headline comparison is not controlled. The baselines likely use different detectors, and the paper doesn't re-run them with Mask3D, so the 15.26% margin over SPAZER could partly reflect detector quality. The abstract says 15.25/14.46% while Section 4.2 says 15.26%; I think they're comparing to different baselines (SeqVLM for Acc@0.25 and SPAZER for Acc@0.5), but the paper should say so explicitly. The hyperparameters (fusion weights α,β,γ, anchor count, rotation unit) are chosen by looking at ScanRefer test performance—Table 4 and Table 6 are effectively test-set sweeps. That inflates the reported numbers. No code or prompts are released, which makes reproduction hard. Also, there are no error bars; with LLM temperature >0, the numbers could swing.

On the stress-test about per-object viewpoint optimization: I agree it's a conceptual weakness. Eq. (8) lets each candidate pick its own rotation angle, so the method isn't truly inferring the observer's viewpoint; it's scoring each object under its best-case rotation. That said, I don't think it's fatal. The true object under the actual viewpoint should have a near-perfect alignment for all anchors, so its S_v should be at least as high as any false object's. The score is still informative, just not as tightly coupled to global viewpoint consistency as the paper claims. Worth a footnote, not a rejection.

Verdict: The paper deserves serious peer review. The architecture is interesting, the task is relevant, and the gains—even if inflated—are worth checking. I'd want to see controlled baselines, a validation split for hyperparameters, and code before trusting the SOTA claim. But it's not a desk reject.

Personally, I'd bring it to the reading group and would cite it if the numbers hold up after controlled comparison.

Referee Report

4 major / 5 minor

Summary. The paper proposes TDVR, a training-free framework for zero-shot 3D visual grounding. TDVR builds a semantic 3D scene graph from Mask3D detections, uses an LLM to disambiguate the natural-language query, performs chain-of-thought structured extraction, and then combines four scores—viewpoint-aware directional reasoning, view-based similarity decoupled reasoning, BERT-based category matching, and CLIP-based appearance matching—to select the target object. Experiments on ScanRefer report 70.85% Acc@0.25 and 64.06% Acc@0.5, which the paper claims is a new state of the art for zero-shot methods and also surpasses several fully supervised baselines. Experiments on Sr3D report 70.00% overall and 79.03% on the viewpoint-dependent subset. The paper also provides ablations, anchor-count sensitivity, LLM comparisons, rotation-angle sensitivity, and inference-time comparisons.

Significance. If the reported results are reliable, TDVR would be a substantial advance: a training-free, LLM-driven pipeline that outperforms prior zero-shot 3D visual grounding methods by a large margin and even exceeds several fully supervised methods on ScanRefer. The paper addresses two real challenges—text ambiguity and viewpoint dependence—and the module-level ablations are consistent with the contribution claims. The framework is also relatively fast (10.21 s per query). However, the headline result rests on a per-object viewpoint maximization step and on hyperparameters selected directly on the test benchmark. These issues, together with a possible information leak in the disambiguation stage, require careful scrutiny before the SOTA claim can be accepted.

major comments (4)

[Section 3.5, Eq. (8)] The viewpoint score for each candidate T is S_v(T) = max_{theta in Theta} sum_j Psi(T, C_j, theta). Because the max is taken independently for every candidate, the method does not infer a single observer viewpoint; it asks whether each object can find some yaw at which the spatial constraints hold. For a wrong object, a suitable rotation often exists, so S_v becomes an 'exists a viewpoint' score rather than a consistency score under a common viewpoint. This is load-bearing: the ablation credits VR with a 19.3% gain (Table 3, ID 4 vs. 3), and the headline 64.06% Acc@0.5 depends on this score. Please either enforce a single global theta shared by all candidates, or provide an explicit experiment comparing the per-object max with a global-viewpoint variant. Without such a control, the viewpoint reasoning claim is not supported.
[Section 3.3, Observer-Centric Synergetic Disambiguation] The disambiguation module feeds appearance descriptions 'generated from 2D images' and spatial relationship descriptions derived from the scene graph into the LLM to produce the disambiguated query. At test time the target is unknown; the paper does not state which object crops and which spatial relations are used as inputs. If the target object's own crop and its relations are used to generate the disambiguated query, then the subsequent appearance and spatial matching is circular and the large 23.4% drop when OD is removed (Table 3, ID 6 vs. 1) is partly explained by target leakage. Please clarify the exact inputs to the disambiguation LLM: are descriptions generated for all candidate objects, or only for the ground-truth target? If the latter, this is a serious methodological flaw that invalidates the zero-shot evaluation.
[Section 4.1 and Tables 3, 4, 6] The fusion weights (alpha=5, beta=3, gamma=1), the number of anchors (N_anc=5), and the rotation unit (10 degrees) are all selected by evaluating final accuracy on the ScanRefer test benchmark. Table 4 shows a peak at N_anc=5 and Table 6 shows monotonic improvement at 10 degrees, but this is test-set tuning, not validation. For a zero-shot method this still constitutes test-set leakage and can inflate the reported improvements. Please report performance on a held-out validation split (or cross-validation) and then fix hyperparameters before reporting test numbers, or provide a sensitivity analysis with multiple seeds and confidence intervals.
[Abstract, Section 4.2, Table 1] The improvement numbers are internally inconsistent: the abstract reports 15.25% and 14.46% improvements, while Section 4.2 states 15.26% over SPAZER on Acc@0.5 and 15.53% on the Multiple subset. More importantly, the comparison in Table 1 is not detector-controlled: the zero-shot baselines use different detectors and backbones, and the text claims 'fair comparison across zero-shot baselines' without supporting evidence. For a credible SOTA claim, please provide a controlled comparison in which all methods use the same detector (e.g., Mask3D) and the same object candidates, or at least report per-method detector details and discuss how this affects the comparison.

minor comments (5)

[Table 1 caption] Typo: 'Comparative results on on the ScanRefer dataset' should read 'on the ScanRefer dataset'.
[Section 3.3] The notation for the number of anchors is inconsistent: the text introduces 'n anchor points' and later uses N_anc in Table 4. Please unify the notation and define N_anc precisely.
[Section 3.5] The rotation center p_rand is said to be randomly sampled, although Eqs. (1)–(4) prove that the relative vectors are independent of the rotation center. Using a random center is therefore unnecessary and could introduce numerical noise; consider fixing the origin or explaining why randomness matters.
[Section 4.1] The paper does not provide the prompts used for GPT-4o and DeepSeek-V3, nor the exact chain-of-thought template or the self-checking procedure. Given that the method is LLM-prompt-driven, this hurts reproducibility. Please include the prompts in the supplementary material or an appendix.
[Section 4.4, Table 7] The inference-time comparison reports only a single number per method. It would be useful to report variance across queries and hardware, and to clarify whether the same detector and LLM are used for all methods.

Circularity Check

2 steps flagged · score 8.0 of 10

The disambiguation module builds its spatial and intra-category descriptions from the target object's true position, so the viewpoint/confusion scores are fitting the answer rather than predicting it.

self definitional [Section 3.3, Observer-Centric Synergetic Disambiguation (target-to-anchor and intra-category description generation)]
"For each anchor, we compute the cosine similarity between the target-to-anchor vector v_tar2anc and eight canonical directional vectors. ... We construct a category-specific local frame centered at the mean coordinates of all same-category objects and determine the target’s relative position. Subsequently, according to the obtained relative position, we generate position descriptions among instances of the same category (e.g., 'Among these [target objects], the one I want is the one located at the [direction]')."

The target object is the unknown output of the grounding task. Using its actual position to derive target-to-anchor vectors and the target's relative position within its category means the disambiguated query encodes the answer before any prediction is made. The later localization is therefore not an inference from the original description but a recovery of information already inserted into the query.
self definitional [Section 3.5, Eq. (7)-(8) and Section 3.6 (S_v and S_c definitions)]
"Ψ(T,C_j,θ)=max_{A_i∈C_j} ReLU( (v_TA_i(θ)·v_ref,j)/||v_TA_i(θ)|| ) ... S_v = max_θ Σ_j Ψ(T,C_j,θ) ... Leveraging the intra-category directional cues parsed in Section 3.4 ..., we evaluate the alignment between the candidate’s displacement and the reference direction R_ref. The confusion score S_c is defined by the cosine similarity: S_c(O_i)= (v_conf·R_ref)/(||v_conf|| ||R_ref||)."

v_ref,j and R_ref are the LLM-parsed forms of the spatial and intra-category descriptions that Section 3.3 generated from the true target's coordinates. Thus S_v and S_c reward a candidate exactly when its geometry matches reference vectors that were constructed from the answer; the max over θ and anchors makes the fit even more permissive. The scores are not independent predictions but reconstructions of the input target position.

full rationale

The paper's headline result is not supported by a self-contained derivation. Section 3.3 constructs the disambiguated query from the target object's actual coordinates (target-to-anchor vectors and the target's relative position within its category). Since the target is the unknown output of grounding, this step injects the answer into the input. Sections 3.5 and 3.6 then define S_v and S_c as alignments against reference vectors that are the LLM-parsed versions of those same target-derived descriptions; the scores therefore correlate with ground truth by construction, not by geometric inference. The large ablation gains assigned to OD (23.4%), VR (19.3%), and VSD (3.2%) are all attributable to modules in this closed loop. The category and appearance scores are independent, but the central SOTA claim rests on the weighted sum that includes the circular scores. No load-bearing self-citation is present; hyperparameter choices (alpha, beta, gamma, rotation unit, anchor count) are tuned on ScanRefer and are an evaluation risk rather than a derivation-level circularity. The per-object max over θ is permissive, but that is a separate correctness concern.

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

No new physical entities, forces, or conserved quantities are introduced. 'Optimal viewpoint set' and 'scene graph' are algorithmic constructs, not invented entities.

free parameters (5)

fusion weights alpha, beta, gamma = alpha=5, beta=3, gamma=1
Set empirically (Section 3.8) with no validation split; tuned against ScanRefer accuracy.
rotation unit angle = 10 degrees
Chosen based on Table 6; 10 degrees gives best ScanRefer Acc@0.5 and is used for all experiments.
number of anchors N_anc = 5
Selected from Table 4; performance peaks at 5 anchors on ScanRefer.
depth consistency threshold epsilon = not numericized ('extremely small threshold')
Used in Algorithm 1 to select 2D crops; no concrete value given, so replication requires guessing.
LLM temperatures = GPT-4o temp=0.7, DeepSeek-V3 temp=0.3
Chosen arbitrarily for generation and parsing; affects randomness and reproducibility.

assumptions (6)

standard math Rotating centroids about any center preserves relative displacement vectors (p'_B - p'_A = R(p_B - p_A))
Section 3.5 Eq. (1)-(3); used to justify viewpoint inference independent of rotation center.
domain assumption Vertical spatial relations are unambiguous and require no pitch optimization
Section 3.5: 'Z-axis-based vertical judgment is unambiguous without pitch optimization'; this is an assumption about ScanRefer/Sr3D descriptions, not proven.
domain assumption Mask3D provides accurate instance segmentation and category labels
Section 4.1: Mask3D is the detector for all candidates; if it misses or mislabels objects, the method cannot ground them.
domain assumption The LLM generates faithful, disambiguated queries and structured relation tuples
Sections 3.3-3.4 rely on GPT-4o/DeepSeek-V3 to fuse descriptions and parse binary spatial tuples; errors propagate to all scores.
domain assumption CLIP and BERT embeddings reflect semantic similarity between object crops/text and query categories
Section 3.7 uses frozen CLIP for appearance and BERT for category; if embeddings misalign, S_cat and S_app mislead.
ad hoc to paper The eight canonical directional vectors correctly encode spatial prepositions
Section 3.3 selects direction by max cosine similarity to eight canonical vectors; the specific vectors are not enumerated, so this mapping is a design choice.

how reviews work

0 comments

Cite this review

Pith. "Pith review of TDVR: Joint Text Disambiguation and Viewpoint Reasoning for Zero-Shot 3D Visual Grounding." pith.science (2026). https://pith.science/paper/Z7XHHHNS

@misc{pith2026260803763,
  author       = {Pith},
  title        = {Pith review of: TDVR: Joint Text Disambiguation and Viewpoint Reasoning for Zero-Shot 3D Visual Grounding},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/Z7XHHHNS}},
  note         = {Machine review of arXiv:2608.03763}
}

read the original abstract

Zero-shot 3D visual grounding aims to localize specific objects based on textual descriptions and 3D visual input. However, the effectiveness of existing methods is significantly hindered by the ambiguous query text and deficient viewpoints. To address these issues, we propose TDVR, a training-free reasoning framework that disambiguates the input text and infers accurate viewpoints for zero-shot 3D visual grounding. First, we construct semantic 3D scene graph from the detected instances in the 3D point cloud. Subsequently, we put the original query, appearance and spatial relationship descriptions into the LLM for fusion, thereby disambiguating the initial input. We leverage chain-of-thought reasoning to generate the structured representation of disambiguated query. Then taking the scene graph and structured query as input, we get the optimal view via viewpoint reasoning to solve the problem of missing viewpoints during grounding. Based on the obtained optimal viewpoint, we further discriminate the distracting objects, enabling the model with the ability to distinguish similar instances. After that, we match the category text and appearance images with the query by computing the similarity of feature vectors. Finally, the target object was identified by integrating the viewpoint score, confusion score, category score, and appearance score. Compared with previous methods, our TDVR has stronger capabilities in viewpoint reasoning, similar object discrimination, and ambiguous query understanding. Experimental results on the public ScanRefer dataset show that our method outperforms the existing state-of-the-art methods by 15.25% and 14.46% in Acc@0.25 and Acc@0.5 respectively, demonstrating the effectiveness of our TDVR in addressing ambiguous query text and deficient viewpoints.

Figures

Figures reproduced from arXiv: 2608.03763 by the authors.

**Figure 2.** Overview of TDVR. TDVR begins by refining the raw query via the Observer-Centric Synergetic Disambiguation [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗

**Figure 3.** Schematic diagram of the Viewpoint-Aware Direc [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗

Figures from the paper (2 more)

**Figure 4.** Figure 4: Schematic diagram of the View-based Similarity [PITH_FULL_IMAGE:figures/full_fig_p005_4.png]

**Figure 5.** Figure 5: Qualitative comparison of 3D visual grounding results. We visualize four challenging cases (a–d) that involve spatial [PITH_FULL_IMAGE:figures/full_fig_p007_5.png]

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

47 extracted references · 22 canonical work pages

[1]

Panos Achlioptas, Ahmed Abdelreheem, Fei Xia, Mohamed Elhoseiny, and Leonidas Guibas. 2020. Referit3d: Neural listeners for fine-grained 3d object identification in real-world scenes. InEuropean conference on computer vision. Springer, 422–440

2020
[2]

Haonan Chang, Kowndinya Boyalakuntla, Shiyang Lu, Siwei Cai, Eric Jing, Shreesh Keskar, Shijie Geng, Adeeb Abbas, Lifeng Zhou, Kostas Bekris, et al

work page
[3]

Dave Zhenyu Chen, Angel X Chang, and Matthias Nießner. 2020. Scanrefer: 3d object localization in rgb-d scans using natural language. InEuropean conference on computer vision. Springer, 202–221

2020
[4]

Keqin Chen, Zhao Zhang, Weili Zeng, Richong Zhang, Feng Zhu, and Rui Zhao
[5]

Angela Dai, Angel X Chang, Manolis Savva, Maciej Halber, Thomas Funkhouser, and Matthias Nießner. 2017. Scannet: Richly-annotated 3d reconstructions of indoor scenes. InProceedings of the IEEE conference on computer vision and pattern recognition. 5828–5839

2017
[6]

Shikra: Unleashing multimodal llm’s referential dialogue magic.arXiv preprint arXiv:2306.15195(2023)

arXiv 2023
[7]

Chun Feng, Joy Hsu, Weiyu Liu, and Jiajun Wu. 2024. Naturally supervised 3d visual grounding with language-regularized concept learners. InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. 13269–13278

work page 2024
[8]

Jacob Devlin, Ming-Wei Chang, Kenton Lee, and Kristina Toutanova. 2019. Bert: Pre-training of deep bidirectional transformers for language understanding. In Proceedings of the 2019 conference of the North American chapter of the association for computational linguistics: human language technologies, volume 1 (long and short papers). 4171–4186

2019

Show all 47 references

[9]

Liang Geng, Jianqin Yin, Gang Chen, and Qingxuan Jia. 2025. Pseudo-EV: Enhanc- ing 3D Visual Grounding with Pseudo Embodied Viewpoint.IEEE Transactions on Circuits and Systems for Video Technology(2025)

2025
[10]

Liang Geng and Jianqin Yin. 2024. Viewinfer3d: 3d visual grounding based on embodied viewpoint inference.IEEE Robotics and Automation Letters9, 9 (2024), 7469–7476

2024
[11]

Qiao Gu, Ali Kuwajerwala, Sacha Morin, Krishna Murthy Jatavallabhula, Bipasha Sen, Aditya Agarwal, Corban Rivera, William Paul, Kirsty Ellis, Rama Chellappa, et al. 2024. Conceptgraphs: Open-vocabulary 3d scene graphs for perception and planning. In2024 IEEE International Conf...

2024
[12]

Hongjie Gu, Jinlong Fan, Liang Zheng, Jing Zhang, and Yuxiang Yang. 2025. DDPA-3DVG: vision-language dual-decoupling and progressive alignment for 3D visual grounding. InProceedings of the Thirty-Fourth International Joint Conference on Artificial Intelligence. 1044–1052

2025
[13]

Namgyu Ho, Laura Schmid, and Se-Young Yun. 2023. Large language models are reasoning teachers. InProceedings of the 61st annual meeting of the association for computational linguistics (volume 1: long papers). 14852–14882

2023
[14]

Wenxuan Guo, Xiuwei Xu, Ziwei Wang, Jianjiang Feng, Jie Zhou, and Jiwen Lu. 2025. Text-guided sparse voxel pruning for efficient 3d visual grounding. In Proceedings of the Computer Vision and Pattern Recognition Conference. 3666–3675

2025
[15]

Takeshi Kojima, Shixiang Shane Gu, Machel Reid, Yutaka Matsuo, and Yusuke Iwasawa. 2022. Large language models are zero-shot reasoners.Advances in neural information processing systems35 (2022), 22199–22213

2022
[16]

Zhao Jin, Rong-Cheng Tu, Jingyi Liao, Wenhao Sun, Xiao Luo, Shunyu Liu, and Dacheng Tao. 2025. SPAZER: Spatial-Semantic Progressive Reasoning Agent for Zero-shot 3D Visual Grounding.arXiv preprint arXiv:2506.21924(2025)

2025 arXiv
[17]

Rong Li, Shijie Li, Lingdong Kong, Xulei Yang, and Junwei Liang. 2025. Seeground: See and ground for zero-shot open-vocabulary 3d visual grounding. InProceedings of the Computer Vision and Pattern Recognition Conference. 3707–3717

2025
[18]

Fuhao Li, Huan Jin, Bin Gao, Liaoyuan Fan, Lihui Jiang, and Long Zeng. 2025. Nugrounding: A multi-view 3d visual grounding framework in autonomous driving.arXiv preprint arXiv:2503.22436(2025)

2025 arXiv
[19]

Jiawen Lin, Shiran Bian, Yihang Zhu, Wenbin Tan, Yachao Zhang, Yuan Xie, and Yanyun Qu. 2025. SeqVLM: Proposal-Guided Multi-View Sequences Reasoning via VLM for Zero-Shot 3D Visual Grounding. InProceedings of the 33rd ACM International Conference on Multimedia. 3094–3103

2025
[20]

Fangjian Lin, Jianlong Yuan, Sitong Wu, Fan Wang, and Zhibin Wang. 2023. Uninext: Exploring a unified architecture for vision recognition. InProceedings of the 31st ACM International Conference on Multimedia. 3200–3208

2023
[21]

Yuanyuan Liu, Haiyang Mei, Dongyang Zhan, Jiayue Zhao, Dongsheng Zhou, Bo Dong, and Xin Yang. 2026. View-on-Graph: Zero-Shot 3D Visual Grounding via Vision-Language Reasoning on Scene Graphs. InProceedings of the AAAI Conference on Artificial Intelligence, Vol. 40. 7386–7394

2026
[22]

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

2024 arXiv
[23]

Pan Lu, Liang Qiu, Wenhao Yu, Sean Welleck, and Kai-Wei Chang. 2023. A survey of deep learning for mathematical reasoning. InProceedings of the 61st annual meeting of the association for computational linguistics (volume 1: long papers). 14605–14631

2023
[24]

Zhenyang Liu, Yikai Wang, Sixiao Zheng, Tongying Pan, Longfei Liang, Yanwei Fu, and Xiangyang Xue. 2025. Reasongrounder: Lvlm-guided hierarchical feature splatting for open-vocabulary 3d visual grounding and reasoning. InProceedings of the Computer Vision and Pattern Recogniti...

2025
[25]

Chuofan Ma, Yi Jiang, Jiannan Wu, Zehuan Yuan, and Xiaojuan Qi. 2024. Groma: Localized visual tokenization for grounding multimodal large language models. InEuropean Conference on Computer Vision. Springer, 417–435

2024
[26]

Ziyang Lu, Yunqiang Pei, Guoqing Wang, Peiwei Li, Yang Yang, Yinjie Lei, and Heng Tao Shen. 2024. Scaneru: Interactive 3d visual grounding based on embodied reference understanding. InProceedings of the AAAI Conference on Artificial Intelligence, Vol. 38. 3936–3944

2024
[27]

Boyu Mi, Hanqing Wang, Tai Wang, Yilun Chen, and Jiangmiao Pang. 2025. Language-to-Space Programming for Training-Free 3D Visual Grounding. In Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing. 3844–3864

2025
[28]

Yunze Man, De-An Huang, Guilin Liu, Shiwei Sheng, Shilong Liu, Liang-Yan Gui, Jan Kautz, Yu-Xiong Wang, and Zhiding Yu. 2025. Argus: Vision-centric reasoning with grounded chain-of-thought. InProceedings of the Computer Vision and Pattern Recognition Conference. 14268–14280

2025
[29]

Hanoona Rasheed, Muhammad Maaz, Sahal Shaji, Abdelrahman Shaker, Salman Khan, Hisham Cholakkal, Rao M Anwer, Eric Xing, Ming-Hsuan Yang, and Fahad S Khan. 2024. Glamm: Pixel grounding large multimodal model. InPro- ceedings of the IEEE/CVF Conference on Computer Vision and Pat...

2024
[30]

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

2021
[31]

Jonas Schult, Francis Engelmann, Alexander Hermans, Or Litany, Siyu Tang, and Bastian Leibe. 2023. Mask3d: Mask transformer for 3d semantic instance segmentation. In2023 IEEE International Conference on Robotics and Automation (ICRA). IEEE, 8216–8223

2023
[32]

Antoni Rosinol, Arjun Gupta, Marcus Abate, Jingnan Shi, and Luca Carlone. 2020. 3D dynamic scene graphs: Actionable spatial perception with places, objects, and humans.arXiv preprint arXiv:2002.06289(2020)

2020 arXiv
[33]

Shihao Wang, Zhiding Yu, Xiaohui Jiang, Shiyi Lan, Min Shi, Nadine Chang, Jan Kautz, Ying Li, and Jose M Alvarez. 2025. Omnidrive: A holistic vision-language dataset for autonomous driving with counterfactual reasoning. InProceedings of the computer vision and pattern recognit...

2025
[34]

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

2024
[35]

Xinyi Wang, Na Zhao, Zhiyuan Han, Dan Guo, and Xun Yang. 2025. Augre- fer: Advancing 3d visual grounding via cross-modal augmentation and spatial relation-based referring. InProceedings of the AAAI Conference on Artificial Intel- ligence, Vol. 39. 8006–8014

2025
[36]

Xuezhi Wang, Jason Wei, Dale Schuurmans, Quoc Le, Ed Chi, Sharan Narang, Aakanksha Chowdhery, and Denny Zhou. 2022. Self-consistency improves chain of thought reasoning in language models.arXiv preprint arXiv:2203.11171(2022)

2022 arXiv
[37]

Jason Wei, Xuezhi Wang, Dale Schuurmans, Maarten Bosma, Fei Xia, Ed Chi, Quoc V Le, Denny Zhou, et al. 2022. Chain-of-thought prompting elicits reasoning in large language models.Advances in neural information processing systems35 (2022), 24824–24837

2022
[38]

Yuan Wang, Ya-Li Li, WU Eastman ZY, and Shengjin Wang. 2025. LIBA: Lan- guage instructed multi-granularity bridge assistant for 3D visual grounding. In Proceedings of the AAAI Conference on Artificial Intelligence, Vol. 39. 8114–8122

2025
[39]

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

2024
[40]

Shun-Cheng Wu, Johanna Wald, Keisuke Tateno, Nassir Navab, and Federico Tombari. 2021. Scenegraphfusion: Incremental 3d scene graph prediction from rgb-d sequences. InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. 7515–7525

2021
[41]

Hang Yin, Xiuwei Xu, Zhenyu Wu, Jie Zhou, and Jiwen Lu. 2024. Sg-nav: Online 3d scene graph prompting for llm-based zero-shot object navigation.Advances in neural information processing systems37 (2024), 5285–5307

2024
[42]

Runsen Xu, Zhiwei Huang, Tai Wang, Yilun Chen, Jiangmiao Pang, and Dahua Lin. 2024. Vlm-grounder: A vlm agent for zero-shot 3d visual grounding.arXiv preprint arXiv:2410.13860(2024)

2024 arXiv
[43]

Tatiana Zemskova and Dmitry Yudin. 2025. 3dgraphllm: Combining semantic graphs and large language models for 3d scene understanding. InProceedings of the IEEE/CVF International Conference on Computer Vision. 8885–8895

2025
[44]

Zhihao Yuan, Jinke Ren, Chun-Mei Feng, Hengshuang Zhao, Shuguang Cui, and Zhen Li. 2024. Visual programming for zero-shot open-vocabulary 3d visual grounding. InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. 20623–20633. MM ’26, November 10–...

2024
[45]

Wayne Xin Zhao, Kun Zhou, Junyi Li, Tianyi Tang, Xiaolei Wang, Yupeng Hou, Yingqian Min, Beichen Zhang, Junjie Zhang, Zican Dong, et al. 2023. A survey of large language models.arXiv preprint arXiv:2303.182231, 2 (2023), 1–124

2023 arXiv
[46]

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

2024
[2023]

arXiv preprint arXiv:2309.15940(2023)

Context-aware entity grounding with open-vocabulary 3d scene graphs. arXiv preprint arXiv:2309.15940(2023)

2023 arXiv

Pith tools

Reviewed August 5, 2026 · model on record in the stance chip above.

[1] [1]

Panos Achlioptas, Ahmed Abdelreheem, Fei Xia, Mohamed Elhoseiny, and Leonidas Guibas. 2020. Referit3d: Neural listeners for fine-grained 3d object identification in real-world scenes. InEuropean conference on computer vision. Springer, 422–440

2020

[2] [2]

Haonan Chang, Kowndinya Boyalakuntla, Shiyang Lu, Siwei Cai, Eric Jing, Shreesh Keskar, Shijie Geng, Adeeb Abbas, Lifeng Zhou, Kostas Bekris, et al

work page

[3] [3]

Dave Zhenyu Chen, Angel X Chang, and Matthias Nießner. 2020. Scanrefer: 3d object localization in rgb-d scans using natural language. InEuropean conference on computer vision. Springer, 202–221

2020

[4] [4]

Keqin Chen, Zhao Zhang, Weili Zeng, Richong Zhang, Feng Zhu, and Rui Zhao

[5] [5]

Angela Dai, Angel X Chang, Manolis Savva, Maciej Halber, Thomas Funkhouser, and Matthias Nießner. 2017. Scannet: Richly-annotated 3d reconstructions of indoor scenes. InProceedings of the IEEE conference on computer vision and pattern recognition. 5828–5839

2017

[6] [6]

Shikra: Unleashing multimodal llm’s referential dialogue magic.arXiv preprint arXiv:2306.15195(2023)

arXiv 2023

[7] [7]

Chun Feng, Joy Hsu, Weiyu Liu, and Jiajun Wu. 2024. Naturally supervised 3d visual grounding with language-regularized concept learners. InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. 13269–13278

work page 2024

[8] [8]

Jacob Devlin, Ming-Wei Chang, Kenton Lee, and Kristina Toutanova. 2019. Bert: Pre-training of deep bidirectional transformers for language understanding. In Proceedings of the 2019 conference of the North American chapter of the association for computational linguistics: human language technologies, volume 1 (long and short papers). 4171–4186

2019