Pith. sign in

REVIEW 4 major objections 6 minor 49 references

SHREC 2025: Retrieval of Optimal Objects for Multi-modal Enhanced Language and Spatial Assistance (ROOMELSA)

T0 review · 4 major / 6 minor · reviewed 2026-08-05 · deepseek-v4-flash

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

Pith's one-line read ROOMELSA establishes that coarse 3D object retrieval is solved while rank-1 fine-grained matching remains the open problem.

desk verdict ROOMELSA is a genuinely new benchmark and a real resource, but its empirical claims are undercut by human-selected ground-truth proxies and an author-participant overlap that the paper doesn't disclose. read the letter →

arxiv 2508.08781 v1 pith:A4BJ47AI submitted 2025-08-12 cs.CV

Trong-Thuan Nguyen , Viet-Tham Huynh , Quang-Thuc Nguyen , Hoang-Phuc Nguyen , Long Le Bao , Thai Hoang Minh , Minh Nguyen Anh , Thang Nguyen Tien
show 25 more authors
Phat Nguyen Thuan Huy Nguyen Phong Bao Huynh Thai Vinh-Tiep Nguyen Duc-Vu Nguyen Phu-Hoa Pham Minh-Huy Le-Hoang Nguyen-Khang Le Minh-Chinh Nguyen Minh-Quan Ho Ngoc-Long Tran Hien-Long Le-Hoang Man-Khoi Tran Anh-Duong Tran Kim Nguyen Quan Nguyen Hung Dat Phan Thanh Hoang Tran Van Tien Huynh Viet Nhan Nguyen Viet Thien Dinh-Khoi Vo Van-Loc Nguyen Trung-Nghia Le Tam V. Nguyen Minh-Triet Tran
This is my paper · ORCID
classification cs.CV
keywords 3Dobjectretrievalbenchmarkdatasetmask-conditionedlanguagegroundingpanoramicsceneunderstandingCADgalleryfine-grainedrankingSHRECchallenge
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

ROOMELSA is a new benchmark that asks a system to retrieve the exact 3D furniture mesh for a masked object in a panoramic room, given only a free-form text description. The dataset contains 44,445 verified (mask, sentence, 3D model) triples from 1,622 apartments and 5,197 rooms, rendered from 3D-FRONT layouts and 3D-FUTURE meshes. Across the 18 participating SHREC 2025 teams, the top five all achieve perfect Recall@5 and @10 on the 50-query private test, meaning the right mesh is always in the shortlist; differences in Mean Reciprocal Rank (best 0.97, fifth 0.93) come entirely from whether the correct mesh is ranked first. The paper argues this separates coarse retrieval, which is essentially solved, from fine-grained tie-breaking over materials, part structure, and contextual fit, which remains the open problem.

What carries the argument

The central object is the benchmark task itself: a query triple of an equirectangular panorama, a binary mask, and an attribute-rich sentence, with the answer being a ranked list from a gallery of roughly 55,000 CAD meshes. The carrying mechanism that emerges from the evaluation is the two-stage retrieval paradigm—a fast embedding-based vector search that guarantees top-k recall, followed by a semantic or geometric reranker that decides the final order. This shortlist-then-rerank structure is what lets near-perfect recall coexist with imperfect MRR.

What would settle it

Take a random sample of ROOMELSA queries and have three independent annotators select the 'correct' gallery mesh for the same masked object. If they frequently disagree on the CAD ID, the ground-truth labels are non-unique, and reported MRR values partly reflect which proxy was chosen rather than retrieval skill. A second check: measure whether MRR degrades more than R@10 when masks are spatially shifted or sentences are paraphrased; if it does, the benchmark's remaining signal is largely tied to exact phrasing and mask placement.

Watch

Extended reading notes

Core claim

The paper establishes ROOMELSA as the first benchmark explicitly designed for mask-conditioned, language-driven 3D object retrieval, reversing the usual referring-object formulation: the location is given as a mask, and the system must decide which catalog CAD mesh matches the masked object's appearance and semantics. On the private test, coarse retrieval is saturated—every top-five method retrieves the correct mesh within the top ten for all 50 queries. The remaining signal is rank-1 ordering: the winning pipeline (multi-view SIGLIP embeddings, BLIP-2 captions, BGE-M3 similarity) reaches MRR 0.97, while a lightweight frozen-CLIP ensemble with silhouette embeddings and voting reaches 0.93. T

Load-bearing premise

The ground truth for each query is a human-chosen CAD replacement from the 55,000-model gallery, and the benchmark's meaningfulness depends on that replacement being unique, consistent across annotators, and semantically identical to the masked object—properties the paper does not verify.

Editorial extensions

If this is right

  • ROOMELSA separates the field into two measurable abilities: coarse recall, which is saturated at 1.00, and fine-grained rank-1 ordering, which decides leaderboard positions.
  • Improving MRR will require models that distinguish materials, part structures, and contextual/functional fit, not just category or silhouette.
  • A lightweight ensemble of frozen CLIP embeddings plus silhouette cues can reach MRR 0.93, so expensive multimodal training is not required for strong shortlisting.
  • The two-stage architecture (vector shortlist, then reranker) provides a concrete blueprint for deploying language-driven 3D retrieval in AR and robotics.
  • Enriching the gallery with material and part-level variants, or reducing the allowed top-k, would make the benchmark more discriminative for future rounds.

Reading between the lines

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

  • Because ground-truth CAD IDs are human-selected replacements from the gallery rather than the scene's original mesh, the benchmark's scores depend on annotator consistency; a natural extension is measuring inter-annotator agreement on CAD ID choice, which the paper does not report.
  • The saturation of R@10 at 1.00 across all top teams suggests the benchmark's discriminative power currently lives almost entirely in rank-1 ordering; shrinking the shortlist or adding near-duplicate distractors would make the evaluation harder and the leaderboard spreads wider.
  • The success of caption-based reranking implies that attribute-level or part-aware captions for gallery meshes might close the remaining gap more effectively than single-sentence captions.
  • A direct stress test would be to corrupt the mask or paraphrase queries and measure whether MRR drops faster than R@10, revealing how much of the tie-breaking depends on exact linguistic phrasing.
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

4 major / 6 minor

Summary. The paper presents ROOMELSA, a SHREC 2025 benchmark for mask-conditioned, language-driven 3D object retrieval. The task is: given a rendered panoramic room image, a binary mask highlighting an object, and a natural-language description, retrieve the correct CAD mesh from a gallery of 55,000 models. The dataset is built from 3D-FRONT/3D-Future assets, contains 44,445 verified (mask, sentence, mesh) triples over 1,622 apartments / 5,197 rooms, and is split into a public test (44,445 queries) and a private test (50 queries). The authors analyze the top five submissions, report that coarse retrieval is saturated (R@5=R@10=1.00 for all five), and identify fine-grained rank-1 ordering as the remaining bottleneck, with the best MRR at 0.97. They also describe the five winning pipelines in detail, emphasizing two-stage retrieval (fast vector search plus semantic reranking) as the dominant paradigm.

Significance. If the benchmark is sound, ROOMELSA fills a useful gap: existing 3D grounding datasets focus on localization (boxes or clusters), whereas ROOMELSA directly evaluates retrieval of a usable CAD asset from a gallery, conditioned on a mask and free-form language. The dataset is publicly released, the rendering pipeline (BlenderProc over 3D-FRONT/3D-Future) is reproducible, and the metric definitions are clear. The two-stage retrieval insight and the observation that rank-1 ordering, not coarse recall, is the bottleneck are likely to be valuable to the community. However, the empirical conclusions rest on a small, non-independent private test with no error bars, and the ground-truth annotation procedure allows replacement of the original scene mesh by a human-selected gallery proxy, which directly affects what the metrics actually measure. These issues substantially temper the strength of the claims as presented.

major comments (4)
  1. [§4.2 (Stage 2) and §4.1] The ground-truth labels are not necessarily the actual objects in the scenes. Because every panorama is rendered from 3D-FRONT layouts with known 3D-Future mesh IDs, the 'provisional CAD ID' is the exact rendered instance. Yet Stage 2 says annotators 'replace the provisional CAD ID with a more accurate match selected from a gallery of 55,000 models.' This means the target of retrieval is a human-chosen gallery proxy, not the mesh visible in the scene. The reported inter-annotator IoU of 0.87±0.04 measures mask agreement, not uniqueness or identity of the selected mesh. The paper never reports how often replacement occurred, how far the proxy deviates from the rendered instance, or whether multiple plausible proxies exist. Since every metric (R@k, MRR) is computed against this label, the central claim of retrieving the 'exact CAD mesh' is not supported. Please either report replacement st
  2. [§5 and author list] All five teams whose methods are analyzed in detail consist of co-authors of this paper (e.g., Stubborn_Strawberries includes Long Le Bao, Thai Hoang Minh et al.; Ai-Yahh includes Phu-Hoa Pham et al.; BUCCI_GANG, MealsRetrieval, and NoResources similarly overlap with the author list). This contradicts the statement in §5 that 'none of the ROOMELSA organizers participated in the challenge or submitted results.' The manuscript does not disclose this overlap. This is a significant transparency problem for a challenge report: the detailed methods, ablations, and qualitative comparisons are effectively self-authored evaluations. Furthermore, several method-specific constants (the sigmoid coefficients and threshold in Eq. (2), the mask-quality band in §6.3, and the voting weights in §6.5) appear tuned on the public test set, which is fully annotated and explicitly provided for local evaluation
  3. [§7.1 and Table 1] The private test consists of only 50 queries, and the authors correctly note that a single query counts as 0.02 in every metric. Yet the paper draws strong conclusions from small differences: R@1 values of 0.94, 0.92, 0.92, 0.90, and 0.88 correspond to 47, 46, 46, 45, and 44 correct rank-1 results. The abstract's claim that 'only one top-performing model consistently ranked the correct match first' is not supported without confidence intervals or significance tests (e.g., bootstrap or exact binomial pairwise tests). The paper's own analysis shows that the MRR spread between first and fifth place is three queries, yet it treats this as meaningful method-level ranking. Please provide error bars or significance tests, or limit the comparative claims to the well-supported statement that coarse retrieval is saturated and that rank-1 differences are small.
  4. [§4.2 (Stage 1)] The annotation description contains a duplicated sentence and an internal numerical contradiction. The text states: 'The model produces approximately 4,600 drafts per hour, providing complete coverage of all 44,000 masked objects ... The model produces approximately 4,600 drafts per hour, ensuring coverage of the more than 50,000 masked objects referenced in the Introduction.' The Introduction and the rest of the paper report 44,445 query triples; no 'more than 50,000 masked objects' is referenced anywhere. This is a factual inconsistency about the dataset size, which is a central contribution. Please fix the duplicate and clarify the exact number of masked objects covered by the automated drafts and the final verified set.
minor comments (6)
  1. [§4.2] Remove the duplicated sentence in Stage 1; keep only one statement of throughput and coverage.
  2. [§6.1] 'the team achieved the highest overall score in the challenge, establishing the ROOMELSA benchmark' is confusing; the benchmark is not established by the winning team. Rephrase.
  3. [§4.3] The private test is described as 'manually sample 50 diverse scenes, ensuring a balanced distribution across functional room types.' Please specify the exact room-type distribution and how diversity was ensured, as this affects generalizability.
  4. [Eq. (2)] The typesetting of the equation has a missing line break or operator (wadapt appears to be multiplied by a bracket, but the expression is hard to parse). Please reformat.
  5. [§4.1] '1024×512 pixels (internally specified with width = 512×2 and height = 512)' is an awkward way to state the resolution; simplify.
  6. [Global] The paper repeatedly says 'exact CAD mesh that corresponds to the described object' (e.g., §3.2, Fig. 2 caption). In light of the annotation replacement step, these statements should be revised to 'gallery mesh selected by annotators' unless the replacement issue is resolved.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: ROOMELSA is an empirical benchmark report grounded in external 3D-FRONT/3D-Future assets and external model evaluations, with no derivation that reduces to its own inputs.

full rationale

This paper is a benchmark/challenge report rather than a derivation: it constructs a dataset, defines MRR and Recall metrics, and reports participant results. There is no claimed prediction derived from a fitted model parameter, nor any equation in which an output is equivalent to an input by construction. The ground-truth annotation pipeline in Sec. 4.2 says annotators may 'replace the provisional CAD ID with a more accurate match selected from a gallery of 55,000 models'; this is a construct-validity limitation (the target is a human-chosen gallery proxy rather than necessarily the exact rendered mesh), but it is not circular because the benchmark's operational target is simply that human-verified triple, and system outputs are not used to define it. The paper's self-citations (e.g., prior SHREC tracks [13,14]) are contextual and not load-bearing. The most notable issue is that the top-five analyzed teams, as listed in Sec. 5, are composed of the paper's own co-authors, and the same section asserts that 'none of the ROOMELSA organizers participated in the challenge or submitted results.' That factual tension raises concerns about the independence of the evaluation, but it is not a circular derivation: no result is justified by self-citation, and no fitted parameter is renamed as a prediction. Under the specified circularity patterns, the appropriate finding is no significant circularity.

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

The central claim is a benchmark, not a derivation, so the load-bearing assumptions concern the synthetic scene-to-gallery mapping and the curation of ground truth. Method-specific hyperparameters from the five pipelines are listed as free parameters because they affect the reported leaderboard but were not set by an external principle or validated on independent splits.

free parameters (4)
  • wadapt sigmoid coefficients = -10 and 0.13 (Eq. 2)
    Hand-chosen constants in the adaptive geometry weight of COMPASS (Ai-Yahh); they determine when depth and Chamfer similarity are down-weighted and were tuned on the public test set.
  • wCD = not reported
    Weight on Chamfer-distance similarity in Eq. (2); the value is not stated, so COMPASS's final score is not reproducible from the paper.
  • mask quality band = [2200, 100000]
    Threshold in MealsRetrieval to choose text-driven versus image-driven retrieval based on mask quality; no justification or sensitivity analysis is given.
  • ensemble voting weights = +2 for text-then-shape variants, +1 for others
    Hand-assigned weights in NoResources' majority vote; chosen to balance five retrieval strategies without a stated selection criterion.
assumptions (4)
  • domain assumption 3D-FRONT layouts and 3D-Future meshes are a valid proxy for real-world rooms and objects
    The dataset is built entirely by rendering these synthetic assets with BlenderProc, so conclusions about real-world 3D recognition inherit this assumption.
  • domain assumption Human triple-agreement verification yields correct (mask, sentence, model) triples
    Ground truth is asserted via inter-annotator IoU 0.87 and 97% attribute-token retention, but no external validation against objective ground truth is provided.
  • ad hoc to paper A suitable ground-truth mesh exists in the 55,000-model gallery for every mask
    Annotators are allowed to replace the provisional CAD ID with a "more accurate match" from the gallery, so correctness is defined as a human-selected gallery proxy rather than the actual scene object.
  • domain assumption The provided binary mask isolates the target object
    The task assumes the mask is given and sufficiently correct; human refinement only partially corrects boundary leakage, and mask quality is not quantified in the final dataset.

how reviews work

0 comments
Cite this review

Pith. "Pith review of SHREC 2025: Retrieval of Optimal Objects for Multi-modal Enhanced Language and Spatial Assistance (ROOMELSA)." pith.science (2026). https://pith.science/paper/A4BJ47AI

@misc{pith2026250808781,
  author       = {Pith},
  title        = {Pith review of: SHREC 2025: Retrieval of Optimal Objects for Multi-modal Enhanced Language and Spatial Assistance (ROOMELSA)},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/A4BJ47AI}},
  note         = {Machine review of arXiv:2508.08781}
}
read the original abstract

Recent 3D retrieval systems are typically designed for simple, controlled scenarios, such as identifying an object from a cropped image or a brief description. However, real-world scenarios are more complex, often requiring the recognition of an object in a cluttered scene based on a vague, free-form description. To this end, we present ROOMELSA, a new benchmark designed to evaluate a system's ability to interpret natural language. Specifically, ROOMELSA attends to a specific region within a panoramic room image and accurately retrieves the corresponding 3D model from a large database. In addition, ROOMELSA includes over 1,600 apartment scenes, nearly 5,200 rooms, and more than 44,000 targeted queries. Empirically, while coarse object retrieval is largely solved, only one top-performing model consistently ranked the correct match first across nearly all test cases. Notably, a lightweight CLIP-based model also performed well, although it struggled with subtle variations in materials, part structures, and contextual cues, resulting in occasional errors. These findings highlight the importance of tightly integrating visual and language understanding. By bridging the gap between scene-level grounding and fine-grained 3D retrieval, ROOMELSA establishes a new benchmark for advancing robust, real-world 3D recognition systems.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

49 extracted references · 49 canonical work pages

  1. [1]

    Scannet: Richly-annotated 3d reconstructions of indoor scenes

    Dai, A, Chang, AX, Savva, M, Halber, M, Funkhouser, T, Nießner, M. Scannet: Richly-annotated 3d reconstructions of indoor scenes. In: Pro- ceedings of the IEEE conference on computer vision and pattern recogni- tion. 2017, p. 5828–5839

    work page 2017

  2. [2]

    Matterport3d: Learning from rgb-d data in indoor environments

    Chang, A, Dai, A, Funkhouser, T, Halber, M, Niessner, M, Savva, M, et al. Matterport3d: Learning from rgb-d data in indoor environments. arXiv preprint arXiv:170906158 2017

    work page 2017

  3. [3]

    3d-future: 3d furniture shape with texture

    Fu, H, Jia, R, Gao, L, Gong, M, Zhao, B, Maybank, S, et al. 3d-future: 3d furniture shape with texture. International Journal of Computer Vision 2021;129:3313–3337

    work page 2021

  4. [4]

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

    Achlioptas, P, Abdelreheem, A, Xia, F, Elhoseiny, M, Guibas, L. 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. Springer; 2020, p. 422–440

    work page 2020

  5. [5]

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

    Chen, DZ, Chang, AX, Nießner, M. Scanrefer: 3d object localization in rgb-d scans using natural language. In: European conference on computer vision. Springer; 2020, p. 202–221

    work page 2020

  6. [6]

    Vigil3d: A linguistically diverse dataset for 3d visual grounding

    Wang, AT, Gong, Z, Chang, AX. Vigil3d: A linguistically diverse dataset for 3d visual grounding. arXiv preprint arXiv:250101366 2025

    work page 2025

  7. [7]

    2D Image-Based 3D Scene Retrieval

    Abdul-Rashid, H, Yuan, J, Li, B, Lu, Y , Bai, S, Bai, X, et al. 2D Image-Based 3D Scene Retrieval. In: Telea, A, Theoharis, T, Veltkamp, R, editors. Eurographics Workshop on 3D Object Retrieval. 2018,

    work page 2018

  8. [8]

    SHREC’19 track: Extended 2d scene image-based 3D scene retrieval

    Abdul-Rashid, H, Yuan, J, Li, B, Lu, Y , Schreck, T, Bui, NM, et al. SHREC’19 track: Extended 2d scene image-based 3D scene retrieval. Eurographics Workshop on 3D Object Retrieval 2019;700:70

    work page 2019

Show all 49 references
  1. [9]

    SHREC 2019- monocular image based 3D model retrieval

    Li, W, Liu, A, Nie, W, Song, D, Li, Y , Wang, W, et al. SHREC 2019- monocular image based 3D model retrieval. In: Eurographics Workshop 3D Object Retrieval. 2019, p. 1–8

    2019

  2. [10]

    SHREC 2020 track: extended monocular image based 3d model retrieval

    Li, W, Song, D, Liu, A, Nie, W, Zhang, T, Zhao, X, et al. SHREC 2020 track: extended monocular image based 3d model retrieval. In: Eurographics Workshop 3D Object Retrieval. 2020,

    2020

  3. [11]

    SHREC’22 track: Open-set 3D object retrieval

    Feng, Y , Gao, Y , Zhao, X, Guo, Y , Bagewadi, N, Bui, NT, et al. SHREC’22 track: Open-set 3D object retrieval. Computers & Graphics 2022;107:231–240. / Computers & Graphics (2025) 13

    2022

  4. [12]

    Shrec’22 track: Sketch-based 3D shape retrieval in the wild

    Qin, J, Yuan, S, Chen, J, Ben Amor, B, Fang, Y , Hoang-Xuan, N, et al. Shrec’22 track: Sketch-based 3D shape retrieval in the wild. Computers and Graphics 2022

    2022

  5. [13]

    Sketchanimar: Sketch-based 3d animal fine-grained retrieval

    Le, TN, Nguyen, TV , Le, MQ, Nguyen, TT, Huynh, VT, Do, TL, et al. Sketchanimar: Sketch-based 3d animal fine-grained retrieval. Computers & Graphics 2023;116:150–161

    2023

  6. [14]

    Textanimar: text-based 3d animal fine-grained retrieval

    Le, TN, Nguyen, TV , Le, MQ, Nguyen, TT, Huynh, VT, Do, TL, et al. Textanimar: text-based 3d animal fine-grained retrieval. Computers & Graphics 2023;116:162–172

    2023

  7. [15]

    Multi3drefer: Grounding text descrip- tion to multiple 3d objects

    Zhang, Y , Gong, Z, Chang, AX. Multi3drefer: Grounding text descrip- tion to multiple 3d objects. In: Proceedings of the IEEE /CVF Interna- tional Conference on Computer Vision. 2023, p. 15225–15236

    2023

  8. [16]

    Referitgame: Re- ferring to objects in photographs of natural scenes

    Kazemzadeh, S, Ordonez, V , Matten, M, Berg, T. Referitgame: Re- ferring to objects in photographs of natural scenes. In: Proceedings of the 2014 conference on empirical methods in natural language processing (EMNLP). 2014, p. 787–798

    2014

  9. [17]

    Generation and comprehension of unambiguous object descriptions

    Mao, J, Huang, J, Toshev, A, Camburu, O, Yuille, AL, Murphy, K. Generation and comprehension of unambiguous object descriptions. In: Proceedings of the IEEE conference on computer vision and pattern recognition. 2016, p. 11–20

    2016

  10. [18]

    Microsoft coco: Common objects in context

    Lin, TY , Maire, M, Belongie, S, Hays, J, Perona, P, Ramanan, D, et al. Microsoft coco: Common objects in context. In: Computer vision– ECCV 2014: 13th European conference, zurich, Switzerland, September 6-12, 2014, proceedings, part v 13. Springer; 2014, p. 740–755

    2014

  11. [19]

    Mat- tnet: Modular attention network for referring expression comprehension

    Yu, L, Lin, Z, Shen, X, Yang, J, Lu, X, Bansal, M, et al. Mat- tnet: Modular attention network for referring expression comprehension. In: Proceedings of the IEEE conference on computer vision and pattern recognition. 2018, p. 1307–1315

    2018

  12. [20]

    Lavt: Language-aware vision transformer for referring image segmentation

    Yang, Z, Wang, J, Tang, Y , Chen, K, Zhao, H, Torr, PH. Lavt: Language-aware vision transformer for referring image segmentation. In: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition. 2022, p. 18155–18165

    2022

  13. [21]

    Language as queries for referring video object segmentation

    Wu, J, Jiang, Y , Sun, P, Yuan, Z, Luo, P. Language as queries for referring video object segmentation. In: Proceedings of the IEEE /CVF Conference on Computer Vision and Pattern Recognition. 2022, p. 4974– 4984

    2022

  14. [22]

    Masked- attention mask transformer for universal image segmentation

    Cheng, B, Misra, I, Schwing, AG, Kirillov, A, Girdhar, R. Masked- attention mask transformer for universal image segmentation. In: Pro- ceedings of the IEEE /CVF conference on computer vision and pattern recognition. 2022, p. 1290–1299

    2022

  15. [23]

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

    Mo, K, Zhu, S, Chang, AX, Yi, L, Tripathi, S, Guibas, LJ, et al. Partnet: A large-scale benchmark for fine-grained and hierarchical part-level 3d object understanding. In: Proceedings of the IEEE /CVF conference on computer vision and pattern recognition. 2019, p. 909–918

    2019

  16. [24]

    Point2cad: Re- verse engineering cad models from 3d point clouds

    Liu, Y , Obukhov, A, Wegner, JD, Schindler, K. Point2cad: Re- verse engineering cad models from 3d point clouds. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. 2024, p. 3763–3772

    2024

  17. [25]

    Mask3d: Mask transformer for 3d semantic instance segmentation

    Schult, J, Engelmann, F, Hermans, A, Litany, O, Tang, S, Leibe, B. Mask3d: Mask transformer for 3d semantic instance segmentation. In: 2023 IEEE International Conference on Robotics and Automation (ICRA). IEEE; 2023, p. 8216–8223

    2023

  18. [26]

    Refmask3d: Language-guided transformer for 3d re- ferring segmentation

    He, S, Ding, H. Refmask3d: Language-guided transformer for 3d re- ferring segmentation. In: Proceedings of the 32nd ACM International Conference on Multimedia. 2024, p. 8316–8325

    2024

  19. [27]

    Segment any 3d object with language

    Lee, S, Zhao, Y , Lee, GH. Segment any 3d object with language. arXiv preprint arXiv:240402157 2024

    2024

  20. [28]

    Learning transferable visual models from natural language supervi- sion

    Radford, A, Kim, JW, Hallacy, C, Ramesh, A, Goh, G, Agarwal, S, et al. Learning transferable visual models from natural language supervi- sion. In: International conference on machine learning. PmLR; 2021, p. 8748–8763

    2021

  21. [29]

    Ulip: Learning a unified representation of language, images, and point clouds for 3d understanding

    Xue, L, Gao, M, Xing, C, Martín-Martín, R, Wu, J, Xiong, C, et al. Ulip: Learning a unified representation of language, images, and point clouds for 3d understanding. In: Proceedings of the IEEE /CVF confer- ence on computer vision and pattern recognition. 2023, p. 1179–1189

    2023

  22. [30]

    Ulip-2: Towards scalable multimodal pre-training for 3d under- standing

    Xue, L, Yu, N, Zhang, S, Panagopoulou, A, Li, J, Martín-Martín, R, et al. Ulip-2: Towards scalable multimodal pre-training for 3d under- standing. In: Proceedings of the IEEE /CVF Conference on Computer Vision and Pattern Recognition. 2024, p. 27091–27101

    2024

  23. [31]

    Objaverse-xl: A universe of 10m + 3d objects

    Deitke, M, Liu, R, Wallingford, M, Ngo, H, Michel, O, Kusupati, A, et al. Objaverse-xl: A universe of 10m + 3d objects. Advances in Neural Information Processing Systems 2023;36:35799–35813

    2023

  24. [32]

    3ur-llm: An end-to- end multimodal large language model for 3d scene understanding

    Xiong, H, Zhuge, Y , Zhu, J, Zhang, L, Lu, H. 3ur-llm: An end-to- end multimodal large language model for 3d scene understanding. arXiv preprint arXiv:250107819 2025

    2025

  25. [33]

    Shrec’14 track: Extended large scale sketch-based 3D shape retrieval

    Li, B, Lu, Y , Li, C, Godil, A, Schreck, T, Aono, M, et al. Shrec’14 track: Extended large scale sketch-based 3D shape retrieval. In: Euro- graphics workshop on 3D object retrieval; vol. 2014. 2014, p. 121–130

    2014

  26. [34]

    Shrec’19 track: Extended 2d scene sketch-based 3D scene retrieval

    Yuan, J, Abdul-Rashid, H, Li, B, Lu, Y , Schreck, T, Bui, NM, et al. Shrec’19 track: Extended 2d scene sketch-based 3D scene retrieval. Eu- rographics Workshop on 3D Object Retrieval 2019;18:70

    2019

  27. [35]

    Llama: Open and e fficient foundation language models

    Touvron, H, Lavril, T, Izacard, G, Martinet, X, Lachaux, MA, Lacroix, T, et al. Llama: Open and e fficient foundation language models. arXiv preprint arXiv:230213971 2023

    2023

  28. [36]

    Sigmoid loss for lan- guage image pre-training

    Zhai, X, Mustafa, B, Kolesnikov, A, Beyer, L. Sigmoid loss for lan- guage image pre-training. In: Proceedings of the IEEE/CVF international conference on computer vision. 2023, p. 11975–11986

    2023

  29. [37]

    Scaling instruction-finetuned language models

    Chung, HW, Hou, L, Longpre, S, Zoph, B, Tay, Y , Fedus, W, et al. Scaling instruction-finetuned language models. Journal of Machine Learning Research 2024;25(70):1–53

    2024

  30. [38]

    C-pack: Packed resources for general chinese embeddings

    Xiao, S, Liu, Z, Zhang, P, Muennigho ff, N, Lian, D, Nie, JY . C-pack: Packed resources for general chinese embeddings. In: Proceedings of the 47th international ACM SIGIR conference on research and development in information retrieval. 2024, p. 641–649

    2024

  31. [39]

    Blip-2: Bootstrapping language-image pre-training with frozen image encoders and large language models

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

    2023

  32. [40]

    Point-bert: Pre- training 3d point cloud transformers with masked point modeling

    Yu, X, Tang, L, Rao, Y , Huang, T, Zhou, J, Lu, J. Point-bert: Pre- training 3d point cloud transformers with masked point modeling. In: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition. 2022, p. 19313–19322

    2022

  33. [41]

    Openshape: Scaling up 3d shape representation towards open-world understand- ing

    Liu, M, Shi, R, Kuang, K, Zhu, Y , Li, X, Han, S, et al. Openshape: Scaling up 3d shape representation towards open-world understand- ing. Advances in neural information processing systems 2023;36:44860– 44879

    2023

  34. [42]

    Panoformer: panorama transformer for indoor 360 ◦ depth estimation

    Shen, Z, Lin, C, Liao, K, Nie, L, Zheng, Z, Zhao, Y . Panoformer: panorama transformer for indoor 360 ◦ depth estimation. In: European Conference on Computer Vision. Springer; 2022, p. 195–211

    2022

  35. [43]

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

    Xiao, B, Wu, H, Xu, W, Dai, X, Hu, H, Lu, Y , et al. Florence-2: Advancing a unified representation for a variety of vision tasks. In: Pro- ceedings of the IEEE /CVF Conference on Computer Vision and Pattern Recognition. 2024, p. 4818–4829

    2024

  36. [44]

    Yolov11: An overview of the key architectural enhancements

    Khanam, R, Hussain, M. Yolov11: An overview of the key architectural enhancements. arXiv preprint arXiv:241017725 2024

    2024

  37. [45]

    Exaone 3.0 7.8 b instruction tuned language model

    An, S, Bae, K, Choi, E, Jungkyu Choi, S, Choi, Y , Hong, S, et al. Exaone 3.0 7.8 b instruction tuned language model. arXiv e-prints 2024;:arXiv–2408

    2024

  38. [46]

    Zero- painter: Training-free layout control for text-to-image synthesis

    Ohanyan, M, Manukyan, H, Wang, Z, Navasardyan, S, Shi, H. Zero- painter: Training-free layout control for text-to-image synthesis. In: Pro- ceedings of the IEEE /CVF Conference on Computer Vision and Pattern Recognition. 2024, p. 8764–8774

    2024

  39. [47]

    Enhanc- ing the reasoning ability of multimodal large language models via mixed preference optimization

    Wang, W, Chen, Z, Wang, W, Cao, Y , Liu, Y , Gao, Z, et al. Enhanc- ing the reasoning ability of multimodal large language models via mixed preference optimization. arXiv preprint arXiv:241110442 2024

    2024

  40. [48]

    Reproducible scaling laws for contrastive language-image learning

    Cherti, M, Beaumont, R, Wightman, R, Wortsman, M, Ilharco, G, Gor- don, C, et al. Reproducible scaling laws for contrastive language-image learning. In: Proceedings of the IEEE /CVF conference on computer vi- sion and pattern recognition. 2023, p. 2818–2829

    2023

  41. [49]

    Expanding performance boundaries of open-source multimodal models with model, data, and test-time scaling

    Chen, Z, Wang, W, Cao, Y , Liu, Y , Gao, Z, Cui, E, et al. Expanding performance boundaries of open-source multimodal models with model, data, and test-time scaling. arXiv preprint arXiv:241205271 2024

    2024

Pith tools

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