Pith. sign in

REVIEW 5 major objections 4 minor 1 cited by

AugRefer: Advancing 3D Visual Grounding via Cross-Modal Augmentation and Spatial Relation-based Referring

T0 review · 5 major / 4 minor · reviewed 2026-08-10 · deepseek-v4-flash

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

Pith's one-line read AugRefer claims that generating synthetic text-3D pairs and injecting spatial relations into decoder attention lifts 3D visual grounding accuracy by up to 9.81 points on Nr3D and 6.58 points on Sr3D over the BUTD-DETR baseline.

desk verdict Useful cross-modal augmentation for 3DVG, but a potential test-set leakage in the augmentation source pool must be resolved before the reported gains can be trusted. read the letter →

arxiv 2501.09428 v1 pith:GAFZ5UYC submitted 2025-01-16 cs.CV

Xinyi Wang , Na Zhao , Zhiyuan Han , Dan Guo , Xun Yang This is my paper · ORCID
classification cs.CV
keywords 3Dvisualgroundingcross-modalaugmentationdataspatialrelationreasoningtransformerdecoderpointcloudreferringexpressioncomprehensionfoundationmodelcaptioning
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

AugRefer addresses two bottlenecks in 3D visual grounding: too few text-3D training pairs and underuse of spatial context. Its cross-modal augmentation inserts new objects into real 3D scenes, renders them at object, local, and scene levels, and uses foundation models to generate and refine descriptions, producing about 22.5k extra training pairs. Its Language-Spatial Adaptive Decoder (LSAD) injects global and pairwise spatial relations directly into transformer attention so the model can tell the target apart from distractors. Across ScanRefer, Nr3D, and Sr3D, AugRefer improves both BUTD-DETR and EDA, including a +9.81 Acc@0.25 gain on Nr3D and +6.58 on Sr3D over BUTD-DETR, reaching state-of-the-art results on ScanRefer and Sr3D. The authors' claim is that the two mechanisms are complementary and together translate into consistent gains for existing 3DVG models.

What carries the argument

The load-bearing mechanism is the Language-Spatial Adaptive Decoder (LSAD), a transformer decoder layer whose attention is augmented with spatial biases. Pairwise spatial relations between object proposals—Euclidean distance plus horizontal and vertical sine/cosine components of the connecting line—are MLP-mapped and added to the attention logits; global spatial relations are the normalized coordinates of the object center in the scene, injected the same way. The companion data mechanism is the cross-modal augmentation pipeline: stander objects are inserted on the ground plane with collision-free placement, rendered from three angles at three granularities (object, local, scene), and captioned by BLIP2 with prompts tailored to each level, then corrected and rephrased by GPT-3.5.

What would settle it

A concrete check: take the augmented pairs and test whether a strong vision-language model can re-locate the inserted object from the generated caption in the rendered image; a low re-grounding hit rate would indicate the captions are not aligned with the inserted object and would undermine the data-side claim. A cheaper quantitative version is to train the baseline on original pairs plus the same number of caption-object pairings randomly shuffled; if accuracy matches the AugRefer gain, the benefit is quantity rather than semantic alignment.

Watch

Extended reading notes

Core claim

The central claim is that 3D visual grounding can be advanced by combining two mechanisms: data-side, synthetic text-3D pairs generated by inserting objects into scenes and captioning the resulting renders with foundation models; and model-side, a decoder that explicitly injects spatial relations into attention. The paper demonstrates the claim by appending both components to two transformer-based baselines and showing accuracy improvements on every one of the three benchmarks. The decoder carries the model-side argument: LSAD applies cross-attention to language first, then pairwise spatial attention over object proposals, then global spatial attention over scene-wide position, with spatial features added to the attention logits. The augmentation carries the data-side argument: multi-angle, multi-level rendering plus BLIP2/GPT-3.5 captioning produces about 22.5k extra pairs, and the authors identify three generated pairs per scene and level as the best working point.

Load-bearing premise

The load-bearing premise is that the captions produced by BLIP2 and GPT-3.5 are accurate and semantically aligned with the inserted object; if they are noisy, the extra pairs mislead training—and the paper's own ablation shows accuracy falling when the number of generated pairs per scene rises from three to five.

Editorial extensions

If this is right

  • Any transformer-based 3DVG model can take the generated text-3D pairs without changing its architecture; training time grows but inference time does not.
  • Explicit spatial attention should help most in 'multiple' splits, where distractors of the same category must be told apart by relation to anchors or scene position.
  • The three caption granularities are complementary: each alone helps less than the combination, so descriptions should mix object attributes, local anchors, and scene-level context.
  • Language-first ordering in the decoder matters: grounding with text before spatial aggregation works better than spatial-first alternatives.
  • Adding too many generated pairs per scene (five instead of three) hurts accuracy, so augmentation quantity should be balanced against caption noise.

Reading between the lines

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

  • Inference: The augmentation recipe appears portable beyond grounding: unlabeled 3D scenes plus an object bank could generate paired data for navigation instruction, spatial QA, or scene-graph tasks, as long as a captioner can reliably describe the inserted object.
  • Inference: The ceiling on augmentation benefit is caption fidelity. A consistency filter—e.g., trying to re-ground each generated caption to the inserted object and dropping failures—would likely let models use more than three pairs per scene before noise dominates.
  • Inference: The spatial-bias-in-attention design suggests a generic recipe: encode relative geometry as MLP-mapped vectors and add to QK^T, which could be applied to other relation-sensitive tasks beyond 3D, such as video grounding or multi-object tracking.
  • Inference: Because the paper only tests two baselines (the strongest open one and an older one), the claim that the pairs are universally usable remains open; testing on the not-open-source strong model or future ones would be the natural next check.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

5 major / 4 minor

Summary. The paper proposes AugRefer, a method for 3D visual grounding with two components: (1) a cross-modal augmentation pipeline that inserts external 3D objects into scenes, renders them at multiple angles and granularities, and uses BLIP2 and GPT-3.5 to generate new text-3D training pairs; and (2) a Language-Spatial Adaptive Decoder (LSAD) that injects global and pairwise spatial relations into cross-modal attention. The method is integrated into BUTD-DETR and EDA and evaluated on ScanRefer, Nr3D, and Sr3D, reporting consistent accuracy gains and SOTA or near-SOTA results, along with ablations in Tables 3-7.

Significance. If the reported results are valid, the cross-modal augmentation is a valuable plug-and-play contribution that can enrich text-3D training data for existing 3DVG methods, and the LSAD decoder provides a concrete way to exploit spatial relations in transformer decoders. The paper's extensive ablations (Tables 3-7) support both components individually and in combination, and the proposed augmentation is clearly described at a conceptual level. The main risk to significance is the unresolved possibility of train/test leakage in the augmentation source pool, which could invalidate the reported gains; this must be addressed before the results can be accepted as evidence for the method.

major comments (5)
  1. [Algorithm 1 (Appendix A) and Sec. 3.1] Algorithm 1 samples the object to insert by 'randomly choosing another scene Sj' from '3D indoor dataset S' (the ScanNet dataset), with no statement that Sj is restricted to the training split. The main text similarly says the external object is 'randomly selected from other scenes.' Since ScanRefer, Nr3D, and Sr3D are all built from ScanNet, this creates a concrete risk that a point cloud from a test scene is inserted into a training scene, giving the model direct access to test-set geometry during training. If this is the case, the reported gains (e.g., +9.81 on Nr3D, +6.58 on Sr3D) are not valid evidence for the method. The authors must specify explicitly that the augmentation source pool is restricted to the training split, and if it is not, they must rerun the experiments with a split-restricted pool and report the new numbers.
  2. [Sec. 4.3, Table 5] The augmentation quantity n is chosen as 3 based on the overall Acc@0.25 values reported in Table 5, which appear to be test-set results on ScanRefer. The text says 'we set n to 3' directly from these results, but there is no statement that this decision was made on a held-out validation split. If Table 5 reports test-set performance, this is test-set tuning of a central hyperparameter, which inflates the reported effectiveness of the augmentation component. The authors should clarify the split used for model selection, or provide validation-based selection.
  3. [Sec. 4.1 and Tables 1-3] All reported numbers are single-run point estimates with no error bars, variance, significance tests, or multiple-seed averages. Since the claimed improvements over the baselines are often in the 2-4 percentage-point range (e.g., +2.10 on ScanRefer, +4.41 on Nr3D), run-to-run variance could be a substantial fraction of these differences. The authors should provide at least three independent runs with mean and standard deviation for the main comparisons, or otherwise justify that the gains are outside training noise.
  4. [Sec. 3.1 and Implementation Details] The augmentation pipeline contains several hyperparameters that are never specified, most notably the value of M used to select the top CLIP images for captioning, and the exact number of rendered images per insertion level. Without these values, the augmentation pipeline is not fully reproducible. Please report all thresholds and counts used in the final experiments.
  5. [Appendix A, Pairwise Spatial Attention] The pairwise spatial relation vector rp_ij is defined only by reference to 'Chen et al. 2022' rather than being specified in the paper. To make the LSAD module self-contained and reproducible, the paper should give the exact formula for the five-dimensional relation vector (distance, sine and cosine of horizontal and vertical angles) in the main text or appendix.
minor comments (4)
  1. [Table 2] The row labeled 'SAT (Roh et al. 2022)' appears to cite the wrong reference: SAT is Yang et al. 2021b, while LanguageRefer is Roh et al. 2022. Please correct the citation.
  2. [Sec. 4.3] The text says 'our cross-model augmentation' where 'cross-modal' is intended; there are also inconsistent capitalizations of 'SR3D' and 'NR3D' across the manuscript.
  3. [Fig. 3 and Sec. 3.1] The description of the multi-angle camera placement states cameras are positioned at 0, 45, and 90 degrees and rotated around the object, but Fig. 3(a) shows only 'Angle1, Angle2, Angle3'; clarifying the exact camera trajectories would help reproducibility.
  4. [References] The reference 'Yang et al. 2024a' is listed as NeurIPS 2023 in the bibliography but '2024a' in the citations; please standardize the year and venue information.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: AugRefer's gains are empirical against held-out benchmarks, and no derivation reduces to its own inputs.

full rationale

The central claims of AugRefer are empirical: cross-modal augmentation and the language-spatial adaptive decoder are evaluated on the ScanRefer, Nr3D, and Sr3D benchmarks against external baselines BUTD-DETR and EDA, with ablations on held-out splits. No equation or fitted parameter is relabeled as a prediction, and no result is defined in terms of the quantity it is said to predict. The only self-citation, Zhao et al. (2022) for 2D-floor collision detection, is a non-load-bearing implementation detail, and the technique is described in the text (erosion on a floor map), so the argument does not rest on an unverified self-citation. The n=3 augmentation quantity is selected by validation, and the paper itself reports degradation at n=5 due to generated-pair noise (Table 5), which is a robustness limitation rather than circularity. One validity concern is flagged: Algorithm 1 samples the inserted object from '3D indoor dataset S' without stating that S is restricted to the training split; if S included evaluation scenes, test geometry could leak into augmented training data. That would undermine the benchmark comparison, but it is a data-hygiene or correctness issue, not a circular definition or fitted-prediction issue, and the manuscript provides no explicit confirmation that leakage occurs.

Assumptions & free parameters 3 free parameters · 3 assumptions · 0 invented entities

The central claims rest on empirical assumptions about caption quality, rendering fidelity, and the usefulness of generated pairs. The paper provides indirect support through ablations, but the noise sensitivity in Table 5 and the absence of error bars leave these assumptions only partially verified.

free parameters (3)
  • n (number of augmented pairs per scene per level) = 3
    Selected via the ablation in Table 5; n=3 gives the best overall accuracy, while n=5 degrades performance due to caption noise.
  • M (top CLIP images selected for captioning) = not reported
    The paper states that top M images are selected but never specifies M, leaving an important threshold for the caption pipeline undefined.
  • Number of inserted objects per scene = 10
    The appendix states ten distinct object insertions at various levels per scene, a hand-chosen quantity affecting the size of the augmented dataset.
assumptions (3)
  • domain assumption Captions generated by BLIP2 and GPT-3.5 from rendered images are accurate enough to describe the inserted object's category, attributes, and spatial relations.
    Invoked in Section 3.1 'Diverse Description Generation'; the ablation in Table 5 indicates this assumption is only partially true, as more pairs introduce noise and degrade performance.
  • domain assumption Multi-angle and multi-level rendering, followed by CLIP filtering, produces 2D views that reliably capture the inserted object's 3D identity, appearance, and placement in the scene.
    The Hybrid Rendering step in Section 3.1 depends on this; the paper notes that missing point clouds and obstructions can degrade image quality, which is why CLIP filtering is used.
  • domain assumption The ScanNet scenes and object categories used for insertion are representative enough that inserting objects from other scenes yields plausible training examples for the target benchmarks.
    The Object Insertion step in Section 3.1 relies on shared indoor object categories and floor plans; the paper does not analyze domain mismatch between ScanNet and ScanRefer/Nr3D/Sr3D scenes.

how reviews work

0 comments
Cite this review

Pith. "Pith review of AugRefer: Advancing 3D Visual Grounding via Cross-Modal Augmentation and Spatial Relation-based Referring." pith.science (2026). https://pith.science/paper/GAFZ5UYC

@misc{pith2026250109428,
  author       = {Pith},
  title        = {Pith review of: AugRefer: Advancing 3D Visual Grounding via Cross-Modal Augmentation and Spatial Relation-based Referring},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/GAFZ5UYC}},
  note         = {Machine review of arXiv:2501.09428}
}
read the original abstract

3D visual grounding (3DVG), which aims to correlate a natural language description with the target object within a 3D scene, is a significant yet challenging task. Despite recent advancements in this domain, existing approaches commonly encounter a shortage: a limited amount and diversity of text3D pairs available for training. Moreover, they fall short in effectively leveraging different contextual clues (e.g., rich spatial relations within the 3D visual space) for grounding. To address these limitations, we propose AugRefer, a novel approach for advancing 3D visual grounding. AugRefer introduces cross-modal augmentation designed to extensively generate diverse text-3D pairs by placing objects into 3D scenes and creating accurate and semantically rich descriptions using foundation models. Notably, the resulting pairs can be utilized by any existing 3DVG methods for enriching their training data. Additionally, AugRefer presents a language-spatial adaptive decoder that effectively adapts the potential referring objects based on the language description and various 3D spatial relations. Extensive experiments on three benchmark datasets clearly validate the effectiveness of AugRefer.

Figures

Figures reproduced from arXiv: 2501.09428 by the authors.

Figure 1
Figure 1. A brief illustration of our proposed AugRefer: 1) [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. The framework overview of AugRefer. It consists of two components: 1) Cross-Modal Augmentation with three steps: ⃝1 Object Insertion → ⃝2 Hybrid Rendering → ⃝3 Caption Generation; and 2) 3D Visual Grounder, where our designed Language-Spatial Adaptive Decoder (LSAD) aims to enable more precise grounding by incorporating 3D spatial relations. In generated text-3D pairs, more complex situations arise. As shown in [PI… view at source ↗
Figure 3
Figure 3. a) Multi-Angle Camera: For each level of the scene, images are captured from multiple angles. b) Multi￾Level Rendering: The scene is rendered at different levels. of training frames. Likewise, 3D Copy-Paste (Ge et al. 2024) inserts virtual objects into real indoor scenes. In this work, we focus on implementing cross-modal augmentation be￾tween text descriptions and 3D scenes. 3 Methodology Consider a 3D indoor scene… view at source ↗
Figures from the paper (7 more)
Figure 4
Figure 4. Figure 4: At the object-level, we instruct the model to provide [PITH_FULL_IMAGE:figures/full_fig_p003_4.png]
Figure 5
Figure 5. Figure 5: Illustrations of a) Language-Spatial Adaptive De [PITH_FULL_IMAGE:figures/full_fig_p004_5.png]
Figure 4
Figure 4. Figure 4: Multi-Level Caption Generation. Conversation process with BLIP2 and ChatGPT for captioning various level rendering images. Both the Local-Level and Scene￾Level captions utilize the same set of prompts. We describe the approach using the Local-Level as an example. ings.…
Figure 6
Figure 6. Figure 6: Qualitative results with ScanRefer descriptions: (a) “ [PITH_FULL_IMAGE:figures/full_fig_p007_6.png]
Figure 7
Figure 7. Figure 7: Qualitative comparison on samples from ScanRefer dataset. GT + Ours Text The backpack is by the desk in the corner. When facing the refrigerator , the square dining table on the right. EDA This whiteboard is on the floor. It is the monitor on the floor closest to the w…
Figure 8
Figure 8. Figure 8: Qualitative comparison on samples from Nr3D dataset [PITH_FULL_IMAGE:figures/full_fig_p011_8.png]
Figure 9
Figure 9. Figure 9: Qualitative comparison on samples from Sr3D dataset [PITH_FULL_IMAGE:figures/full_fig_p012_9.png]

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. 3DResT: A Strong Baseline for Semi-Supervised 3D Referring Expression Segmentation

    cs.CV 2025-04 conditional novelty 6.0 of 10

    3DResT applies teacher-student semi-supervised learning to 3D referring expression segmentation, promoting high-agreement pseudo-labels into the labeled set and dynamically weighting low-agreement ones, and reports a ...

Reference graph

Works this paper leans on

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

  1. [1]

    L.; Almeida, D.; Altenschmidt, J.; Altman, S.; Anadkat, S.; et al

    Achiam, J.; Adler, S.; Agarwal, S.; Ahmad, L.; Akkaya, I.; Aleman, F. L.; Almeida, D.; Altenschmidt, J.; Altman, S.; Anadkat, S.; et al. 2023. Gpt-4 technical report. arXiv preprint arXiv:2303.08774

    arXiv 2023

  2. [2]

    Achlioptas, P.; Abdelreheem, A.; Xia, F.; Elhoseiny, M.; and Guibas, L. 2020. Referit3d: Neural listeners for fine-grained 3d object identification in real-world scenes. In Computer Vision--ECCV 2020: 16th European Conference, Glasgow, UK, August 23--28, 2020, Proceedings, Part I 16, 422--440. Springer

    work page 2020

  3. [3]

    D.; Dhariwal, P.; Neelakantan, A.; Shyam, P.; Sastry, G.; Askell, A.; et al

    Brown, T.; Mann, B.; Ryder, N.; Subbiah, M.; Kaplan, J. D.; Dhariwal, P.; Neelakantan, A.; Shyam, P.; Sastry, G.; Askell, A.; et al. 2020. Language models are few-shot learners. Advances in neural information processing systems

    work page 2020

  4. [4]

    Z.; Chang, A

    Chen, D. Z.; Chang, A. X.; and Nie ner, M. 2020. Scanrefer: 3d object localization in rgb-d scans using natural language. In European conference on computer vision, 202--221. Springer

    work page 2020

  5. [5]

    Chen, S.; Guhur, P.-L.; Tapaswi, M.; Schmid, C.; and Laptev, I. 2022. Language conditioned spatial relation reasoning for 3d object grounding. Advances in neural information processing systems, 35: 20522--20535

    work page 2022

  6. [6]

    X.; Savva, M.; Halber, M.; Funkhouser, T.; and Nie ner, M

    Dai, A.; Chang, A. X.; Savva, M.; Halber, M.; Funkhouser, T.; and Nie ner, M. 2017. Scannet: Richly-annotated 3d reconstructions of indoor scenes. In Proceedings of the IEEE conference on computer vision and pattern recognition, 5828--5839

    2017

  7. [7]

    Ding, R.; Yang, J.; Jiang, L.; and Qi, X. 2022. Doda: Data-oriented sim-to-real domain adaptation for 3d semantic segmentation. In European Conference on Computer Vision, 284--303. Springer

    work page 2022

  8. [8]

    Ding, R.; Yang, J.; Xue, C.; Zhang, W.; Bai, S.; and Qi, X. 2023. Pla: Language-driven open-vocabulary 3d scene understanding. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 7010--7019

    work page 2023

Show all 45 references
  1. [9]

    Feng, M.; Li, Z.; Li, Q.; Zhang, L.; Zhang, X.; Zhu, G.; Zhang, H.; Wang, Y.; and Mian, A. 2021. Free-form description guided 3d visual graph network for object grounding in point cloud. In Proceedings of the IEEE/CVF International Conference on Computer Vision, 3722--3731

    2021

  2. [10]

    Ge, Y.; Yu, H.-X.; Zhao, C.; Guo, Y.; Huang, X.; Ren, L.; Itti, L.; and Wu, J. 2024. 3D Copy-Paste: Physically Plausible Object Insertion for Monocular 3D Detection. Advances in Neural Information Processing Systems, 36

    2024

  3. [11]

    Guo, Z.; Tang, Y.; Zhang, R.; Wang, D.; Wang, Z.; Zhao, B.; and Li, X. 2023. Viewrefer: Grasp the multi-view knowledge for 3d visual grounding. In Proceedings of the IEEE/CVF International Conference on Computer Vision

    2023

  4. [12]

    T.; and Zhang, H

    Han, Y.; Zhao, N.; Chen, W.; Ma, K. T.; and Zhang, H. 2024. Dual-Perspective Knowledge Enrichment for Semi-supervised 3D Object Detection. In Proceedings of the AAAI Conference on Artificial Intelligence

    2024

  5. [13]

    Hong, Y.; Zhen, H.; Chen, P.; Zheng, S.; Du, Y.; Chen, Z.; and Gan, C. 2023. 3d-llm: Injecting the 3d world into large language models. Advances in Neural Information Processing Systems, 36: 20482--20494

    2023

  6. [14]

    Huang, P.-H.; Lee, H.-H.; Chen, H.-T.; and Liu, T.-L. 2021. Text-guided graph neural networks for referring 3d instance segmentation. In Proceedings of the AAAI Conference on Artificial Intelligence, volume 35, 1610--1618

    2021

  7. [15]

    Huang, S.; Chen, Y.; Jia, J.; and Wang, L. 2022. Multi-view transformer for 3d visual grounding. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 15524--15533

    2022

  8. [16]

    Jain, A.; Gkanatsios, N.; Mediratta, I.; and Fragkiadaki, K. 2022. Bottom up top down detection transformers for language grounding in images and point clouds. In European Conference on Computer Vision, 417--433. Springer

    2022

  9. [17]

    Jiao, P.; Zhao, N.; Chen, J.; and Jiang, Y.-G. 2024. Unlocking textual and visual wisdom: Open-vocabulary 3d object detection enhanced by comprehensive guidance from text and image. In European Conference on Computer Vision, 376--392. Springer

    2024

  10. [18]

    Li, J.; Li, D.; Savarese, S.; and Hoi, S. 2023. Blip-2: Bootstrapping language-image pre-training with frozen image encoders and large language models. In International conference on machine learning, 19730--19742. PMLR

    2023

  11. [19]

    Liu, Y.; Ott, M.; Goyal, N.; Du, J.; Joshi, M.; Chen, D.; Levy, O.; Lewis, M.; Zettlemoyer, L.; and Stoyanov, V. 2019. Roberta: A robustly optimized bert pretraining approach. arXiv preprint arXiv:1907.11692

    2019 arXiv

  12. [20]

    Liu, Z.; Zhang, Z.; Cao, Y.; Hu, H.; and Tong, X. 2021. Group-free 3d object detection via transformers. In Proceedings of the IEEE/CVF International Conference on Computer Vision, 2949--2958

    2021

  13. [21]

    Luo, C.; Di, D.; Yang, X.; Ma, Y.; Xue, Z.; Wei, C.; and Liu, Y. 2025. TrAME: Trajectory-Anchored Multi-View Editing for Text-Guided 3D Gaussian Splatting Manipulation. IEEE Transactions on Multimedia

    2025

  14. [22]

    Luo, J.; Fu, J.; Kong, X.; Gao, C.; Ren, H.; Shen, H.; Xia, H.; and Liu, S. 2022. 3d-sps: Single-stage 3d visual grounding via referred point progressive selection. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 16454--16463

    2022

  15. [23]

    Nekrasov, A.; Schult, J.; Litany, O.; Leibe, B.; and Engelmann, F. 2021. Mix3d: Out-of-context data augmentation for 3d scenes. In 2021 international conference on 3d vision (3dv), 116--125. IEEE

    2021

  16. [24]

    Pan, H.; Cao, Y.; Wang, X.; Yang, X.; and Wang, M. 2024. Finding and Editing Multi-Modal Neurons in Pre-Trained Transformers. In Findings of the Association for Computational Linguistics ACL 2024, 1012--1037

    2024

  17. [25]

    R.; Yi, L.; Su, H.; and Guibas, L

    Qi, C. R.; Yi, L.; Su, H.; and Guibas, L. J. 2017. Pointnet++: Deep hierarchical feature learning on point sets in a metric space. Advances in neural information processing systems, 30

    2017

  18. [26]

    W.; Hallacy, C.; Ramesh, A.; Goh, G.; Agarwal, S.; Sastry, G.; Askell, A.; Mishkin, P.; Clark, J.; et al

    Radford, A.; Kim, J. W.; Hallacy, C.; Ramesh, A.; Goh, G.; Agarwal, S.; Sastry, G.; Askell, A.; Mishkin, P.; Clark, J.; et al. 2021. Learning transferable visual models from natural language supervision. In International conference on machine learning, 8748--8763. PMLR

    2021

  19. [27]

    Roh, J.; Desingh, K.; Farhadi, A.; and Fox, D. 2022. Languagerefer: Spatial-language model for 3d visual grounding. In Conference on Robot Learning, 1046--1056. PMLR

    2022

  20. [28]

    Sheng, H.; Cai, S.; Zhao, N.; Deng, B.; Huang, J.; Hua, X.-S.; Zhao, M.-J.; and Lee, G. H. 2022. Rethinking IoU-based optimization for single-stage 3D object detection. In European Conference on Computer Vision, 544--561. Springer

    2022

  21. [29]

    N.; Kaiser, .; and Polosukhin, I

    Vaswani, A.; Shazeer, N.; Parmar, N.; Uszkoreit, J.; Jones, L.; Gomez, A. N.; Kaiser, .; and Polosukhin, I. 2017. Attention is all you need. Advances in neural information processing systems, 30

    2017

  22. [30]

    Wang, Z.; Huang, H.; Zhao, Y.; Li, L.; Cheng, X.; Zhu, Y.; Yin, A.; and Zhao, Z. 2023. Distilling coarse-to-fine semantic matching knowledge for weakly supervised 3d visual grounding. In Proceedings of the IEEE/CVF International Conference on Computer Vision, 2662--2671

    2023

  23. [31]

    Wu, Y.; Cheng, X.; Zhang, R.; Cheng, Z.; and Zhang, J. 2023. Eda: Explicit text-decoupling and dense alignment for 3d visual grounding. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 19231--19242

    2023

  24. [32]

    Yang, L.; Zhang, Z.; Qi, Z.; Xu, Y.; Liu, W.; Shan, Y.; Li, B.; Yang, W.; Li, P.; Wang, Y.; et al. 2024 a . Exploiting Contextual Objects and Relations for 3D Visual Grounding. Advances in Neural Information Processing Systems, 36

    2024

  25. [33]

    Yang, X.; Feng, F.; Ji, W.; Wang, M.; and Chua, T.-S. 2021 a . Deconfounded video moment retrieval with causal intervention. In Proceedings of the 44th International ACM SIGIR Conference on Research and Development in Information Retrieval, 1--10

    2021

  26. [34]

    Yang, X.; Wang, S.; Dong, J.; Dong, J.; Wang, M.; and Chua, T.-S. 2022. Video moment retrieval with cross-modal neural architecture search. IEEE Transactions on Image Processing, 31: 1204--1216

    2022

  27. [35]

    Yang, X.; Zeng, J.; Guo, D.; Wang, S.; Dong, J.; and Wang, M. 2024 b . Robust Video Question Answering via Contrastive Cross-Modality Representation Learning. SCIENCE CHINA Information Sciences, 67: 1--16

    2024

  28. [36]

    Yang, Z.; Zhang, S.; Wang, L.; and Luo, J. 2021 b . Sat: 2d semantics assisted training for 3d visual grounding. In Proceedings of the IEEE/CVF International Conference on Computer Vision, 1856--1866

    2021

  29. [37]

    Yuan, Z.; Yan, X.; Liao, Y.; Zhang, R.; Wang, S.; Li, Z.; and Cui, S. 2021. Instancerefer: Cooperative holistic understanding for visual grounding on point clouds through instance multi-level contextual referring. In Proceedings of the IEEE/CVF International Conference on Comp...

    2021

  30. [38]

    Zhang, W.; Wang, Z.; and Loy, C. C. 2020. Exploring data augmentation for multi-modality 3d object detection. arXiv preprint arXiv:2012.12741

    2020 arXiv

  31. [39]

    Zhang, Y.; Gong, Z.; and Chang, A. X. 2023. Multi3drefer: Grounding text description to multiple 3d objects. In Proceedings of the IEEE/CVF International Conference on Computer Vision, 15225--15236

    2023

  32. [40]

    Zhao, L.; Cai, D.; Sheng, L.; and Xu, D. 2021. 3DVG-Transformer: Relation modeling for visual grounding on point clouds. In Proceedings of the IEEE/CVF International Conference on Computer Vision, 2928--2937

    2021

  33. [41]

    Zhao, N.; Chua, T.-S.; and Lee, G. H. 2020. Sess: Self-ensembling semi-supervised 3d object detection. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 11079--11087

    2020

  34. [42]

    Zhao, N.; and Lee, G. H. 2022. Static-dynamic co-teaching for class-incremental 3d object detection. In Proceedings of the AAAI Conference on Artificial Intelligence, volume 36, 3436--3445

    2022

  35. [43]

    Zhao, Y.; Zhao, N.; and Lee, G. H. 2022. Synthetic-to-Real Domain Generalized Semantic Segmentation for 3D Indoor Point Clouds. arXiv preprint arXiv:2212.04668

    2022 arXiv

  36. [44]

    , " * write output.state after.block = add.period write newline

    ENTRY address archivePrefix author booktitle chapter edition editor eid eprint howpublished institution isbn journal key month note number organization pages publisher school series title type volume year label extra.label sort.label short.list INTEGERS output.state before.all...

  37. [45]

    write newline

    " write newline "" before.all 'output.state := FUNCTION n.dashify 't := "" t empty not t #1 #1 substring "-" = t #1 #2 substring "--" = not "--" * t #2 global.max substring 't := t #1 #1 substring "-" = "-" * t #2 global.max substring 't := while if t #1 #1 substring * t #2 gl...

Pith tools

Reviewed August 10, 2026 · model on record in the stance chip above.