Pith. sign in

REVIEW 3 major objections 4 minor 47 references

SHREC'25 Track on Multiple Relief Patterns: Report and Analysis

T0 review · 3 major / 4 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 benchmark report claims that segmenting and retrieving multiple relief patterns on 3D meshes is still beyond current algorithms, since no submitted method solved either task on its 1,000-model synthetic dataset.

desk verdict Useful new benchmark, but the 'no existing algorithms' claim outruns the evidence. read the letter →

arxiv 2508.09909 v1 pith:G6BQZOZO submitted 2025-08-13 cs.CG

Gabriele Paolini , Claudio Tortorici , Stefano Berretti , Ahmed Hazem Youssef , Halim Benhabiles , Adnane Cabani , Ruiwen He , Karim Hammoudi
show 5 more authors
Iyyakutti Iyappan Ganapathi Syed Sadaf Ali Divya Velayudhan Maregu Assefa Naoufel Werghi
This is my paper
classification cs.CG MSC 68U0568T45
keywords reliefpatternsgeometrictextures3Dmeshsegmentationsurfaceretrievalbenchmarkdatasetper-facegroundtruthsyntheticmeshesunseen-classgeneralization
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

This paper reports the outcome of a 2025 contest track built around one question: can an algorithm find and isolate repeated geometric textures ("relief patterns") on the surface of a 3D mesh, and then retrieve other meshes carrying the same texture? The track's answer is no, not yet. Across roughly 1,000 procedurally textured synthetic meshes, some carrying two or three different patterns, some with patterns covering only part of the surface, and with training and test pattern classes deliberately disjoint, no submitted method completed either the segmentation or the retrieval task at useful accuracy; the strongest attempt, a multi-view image classifier, reached an AUC of 0.523, essentially random. The paper's contribution is therefore the benchmark itself, with automatic per-face ground truth, plus a diagnosis of why current techniques fail: they are built for single whole-surface patterns, rely on 2D projections that miss occluded textures, and use fixed-radius neighborhoods that blur the boundary between adjacent patterns. If the paper is right, relief pattern analysis is a genuinely open problem with direct applications in artifact restoration, industrial surface inspection, and biomedical imaging.

What carries the argument

The load-bearing object is the dataset itself: roughly 1,000 synthetically generated triangle meshes with procedurally applied relief patterns and automatic per-face ground truth, organized into training (700), retrieval (300), and query (54) sets with deliberately disjoint pattern classes. A relief pattern is defined as a local, repeated deformation on a 3D surface, independent of the object's global shape; the benchmark operationalizes this by applying curated 2D textures as geometric displacements onto base meshes with varied topology. The dataset carries the argument because the track's negative result, that no submitted algorithm solved either task, is meaningful only if the data is a f

What would settle it

One experiment would settle the central claim: train a per-face classifier on the 700-model training set using only local geometric features (curvature, normals, shape index over several radii) and evaluate per-face label agreement on the 54 query meshes. High accuracy on the 14 single-pattern queries together with near-chance accuracy on the 40 two-pattern queries would show the open problem is specifically pattern boundaries and multi-pattern disambiguation; near-chance accuracy everywhere would confirm the paper's broader claim. Strong accuracy on all 54 queries under the track protocol wou

Watch

Extended reading notes

Core claim

The paper's central claim is that relief pattern analysis remains unsolved under general conditions: surfaces carrying one to three patterns, covering part or all of the surface, with quasi-regular organic textures, on complex meshes with self-occlusions. To test this, the organizers generated roughly 1,000 synthetic triangle meshes by procedurally applying curated 2D textures to base models, producing per-face ground truth automatically, and split the texture classes so the 700 training meshes share no classes with the 300 retrieval meshes or the 54 queries. Five groups registered; one submitted a partial retrieval-only solution (a multi-view image classifier, AUC 0.523), and one documented

Load-bearing premise

The report's conclusion that relief pattern analysis is an open problem assumes that procedurally textured synthetic meshes, with their noise-free surfaces and automatic labels, are a faithful proxy for real surfaces carrying reliefs; if real scanned surfaces behave differently, the benchmark's difficulty may not transfer to the applications the paper cites.

Editorial extensions

If this is right

  • The released 700-model training set, one of the largest for relief pattern analysis, gives future methods a standardized place to train and be compared; the track's protocol (query set, retrieval set, per-face ground truth) becomes the measuring stick.
  • Projection-based methods that render a few whole-object views cannot reliably solve the task: patterns hidden inside closed or self-occluding geometry are invisible to them, which explains the near-chance AUC of the one submitted retrieval solution.
  • Retrieval techniques from earlier single-pattern tracks do not transfer: they aggregate features over one representative planar patch, which presupposes a single whole-surface texture, so extending them to multi-pattern, partially covered surfaces requires a different localization step.
  • Segmentation at pattern interfaces is a concrete bottleneck: fixed-radius neighborhoods spanning two adjacent patterns produce ambiguous per-face features, so any viable method must handle boundaries explicitly or adapt its receptive field.
  • Because texture classes are disjoint between training and retrieval/query sets, learning-based methods must recognize previously unseen patterns, so the benchmark tests generalization rather than memorization.

Reading between the lines

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

  • I read the near-chance AUC as evidence that the discriminative signal lives in local surface geometry, not whole-object appearance; a cheap test would be a random forest over per-face curvature and shape-index histograms trained on the 700-model set. If it beats the submitted image classifier on single-pattern queries alone, the bottleneck is specifically pattern boundaries and unseen classes, not
  • The synthetic-to-real gap cuts both ways: procedurally textured, noise-free meshes may be easier than scanned artifacts, or real acquisition noise may make the disjoint-class generalization even harder. A small validation set of photogrammetry-scanned engraved objects would calibrate the benchmark's difficulty against the applications the paper cites.
  • The track's failure rate (one partial submission out of five registrations) mirrors an earlier multi-pattern track, suggesting the bottleneck is a structural mismatch between available methods and the multi-pattern, partially covered setting rather than this dataset's idiosyncrasies; a method that predicts per-face pattern class with scale-adaptive neighborhoods would be the natural next entry.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

3 major / 4 minor

Summary. This paper reports on the SHREC 2025 track on retrieval and segmentation of multiple relief patterns on 3D meshes. The track provides a synthetic benchmark of about 1,054 meshes split into a query set (54), a retrieval set (300), and a training set (700 with per-face ground-truth labels), and defines two tasks: retrieving all meshes sharing at least one relief pattern with a query, and segmenting surfaces containing multiple relief patterns. Five groups registered; one group (OMVMLP) submitted partial retrieval results, and one group (KU-3DSeg) provided preliminary qualitative feedback only. The reported OMVMLP performance is near chance (AUC 0.523, mAP 0.218). From this outcome, the paper concludes that no existing algorithms are capable of effectively performing segmentation or retrieval under the conditions of the dataset, and argues that the benchmark is therefore a valuable open-problem testbed.

Significance. If the benchmark is properly validated and the claims appropriately scoped, this is a useful community resource: it is among the largest relief-pattern datasets with automatic per-face ground truth, and the two-task design (segmentation plus retrieval) addresses a realistic gap in existing SHREC tracks. The paper clearly documents dataset construction, task definitions, and evaluation metrics, and the official webpage provides reproducibility resources. However, the paper's central generalization claim is not supported by the evidence it reports, and the only quantitative submission contains a transductive evaluation step. The benchmark could still be a significant contribution after the claims are revised and baseline experiments are added, but in its current form the paper overstates what the track results demonstrate.

major comments (3)
  1. [Section 1 and Section 7] The claim that 'no existing algorithms are capable of effectively performing segmentation or retrieval under the general conditions imposed by the dataset' is load-bearing but unsupported. The evidence is one partial submission (OMVMLP, Table 1) plus non-submission by four registered groups. Non-participation does not test any algorithm, and a single near-random result cannot establish a universal negative about existing algorithms. Since the track provides a 700-mesh training set with per-face ground truth (Section 2), a straightforward supervised baseline trained on those labels is the obvious missing control; without it, the paper's central conclusion should be weakened to 'no submitted method performed satisfactorily on this benchmark' or supported by additional baseline experiments.
  2. [Sections 3.2.1 and 3.2.2] The quantitative evaluation is not a clean generalization test. In Strategy 2, the classifier is fine-tuned on the query set and then evaluated on the same queries' retrieval relevance labels; this is a transductive/leakage step that inflates or at least obscures the meaning of the reported scores. In addition, Table 1 pools Strategies 1 and 2 and says queries with no relevant retrieval mesh were excluded, but the number of excluded queries is not stated. Without a per-strategy breakdown and the exclusion count, the effective test set size and the near-random AUC cannot be properly interpreted. The authors should report these details or recompute the metrics with the leakage removed.
  3. [Section 3.1 and Figures 7-8] KU-3DSeg is described as a proposed method, but no quantitative evaluation is provided, only illustrative segmented outputs. If this method is used to support the claim of task difficulty, the absence of any metric or evaluation protocol is a critical gap. If it is meant only as preliminary feedback, the paper should say so explicitly and avoid presenting it as evidence about the performance of existing algorithms.
minor comments (4)
  1. [Section 3.2.2 / Figure 6] The caption says 'A single ROC curve is sufficient to evaluate both runs,' but the text says the two strategies differ on the query classes with additional patterns. Unless the ROC curve is identical by construction, the paper should explain why one curve represents both runs, and the reference to 'Fig. 1' in the caption appears to be incorrect.
  2. [Section 3.2.2] Table 1 would benefit from reporting the number of queries remaining after exclusion, the number of relevant models per query, and separate columns for Strategy 1 and Strategy 2. This is needed for the reader to assess the statistical weight of AUC 0.523.
  3. [Section 2] The paper says the full dataset consists of 'approximately 1,000 triangular meshes'; the sum of query, retrieval, and training sets is 1,054. It would be clearer to state the exact total or describe the sets as approximately 1,050.
  4. [Section 3.1] Minor typos: 'adjaceny graph' should be 'adjacency graph'; in Section 4, 'descritptions' should be 'descriptions.'

Circularity Check

1 steps flagged · score 2.0 of 10

Benchmark report is mostly self-contained: no load-bearing self-citation chain or definitional equivalence, but OMVMLP Strategy 2 uses test-query fine-tuning, a disclosed circular evaluation step.

  1. fitted input called prediction [Section 3.2.1 (OMVMLP Strategy 2); Section 3.2.2/Table 1]
    "Strategy 2: the classifier is adapted by replacing its original 15-class output layer with a 4-class output layer, which is then fine-tuned on the set of queries. This adapted classifier is then used to estimate the probability that a target model contains one of the four additional relief classes."

    The adapted classifier is fine-tuned on the query set, and the same query set anchors the retrieval evaluation (membership matrices pooled into Table 1). Thus the query side of the evaluation is in-sample by construction: the model has seen the query patterns during fine-tuning. This is a transductive/leakage protocol, so the reported numbers are not a clean out-of-sample test of generalization. The paper discloses the procedure and does not claim success (AUC 0.523 is near chance), so this is a minor circular step in the evidence, not the load-bearing derivation of the open-problem claim.

full rationale

The paper's central claim—that relief-pattern segmentation/retrieval under the track's conditions is still open—is an empirical benchmark statement, not a derivation from fitted parameters or from a self-citation chain. The ground truth is defined by the procedural texture-generation pipeline, independent of any submitted method; no parameter is fitted to the outcome and then renamed a prediction. Self-citations (e.g., [12,14,32]) appear as contextual prior work and are not load-bearing for the open-problem conclusion. The one genuine circular element is OMVMLP Strategy 2, which fine-tunes on the query set and is then evaluated on those same queries; this is transparently reported and, if anything, makes the already-poor results less informative rather than inflating a success claim. Because this step is confined to a single participant's disclosed protocol and does not force the central conclusion, the overall circularity is low (2/10).

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

No free parameters are fitted by the track design; participant hyperparameters are not part of the paper's central claim. The main assumptions are domain assumptions about the validity of synthetic data and automatic ground truth, plus a disclosed but unquantified evaluation exclusion. No new physical or metaphysical entities are posited.

assumptions (3)
  • domain assumption Synthetic meshes with automatically applied geometric textures yield ground-truth annotations that are a valid testbed for relief pattern analysis
    Invoked in Section 2 to justify replacing scanned surfaces with synthetic ones; if the procedural pipeline generates patterns or boundaries that differ from real acquisitions, the benchmark's conclusions may not transfer.
  • domain assumption Per-face labels produced during dataset generation define the ground truth for both segmentation and retrieval
    The paper assigns relevance per face automatically in Section 1.2; this assumes the texture application pipeline unambiguously determines pattern classes and their spatial boundaries.
  • domain assumption Excluding queries with no relevant retrieval mesh yields interpretable metrics
    Section 3.2.2 removes those queries before computing metrics; the number of excluded queries is not reported, so the representativeness of the reported aggregate values is unknown.

how reviews work

0 comments
Cite this review

Pith. "Pith review of SHREC'25 Track on Multiple Relief Patterns: Report and Analysis." pith.science (2026). https://pith.science/paper/G6BQZOZO

@misc{pith2026250809909,
  author       = {Pith},
  title        = {Pith review of: SHREC'25 Track on Multiple Relief Patterns: Report and Analysis},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/G6BQZOZO}},
  note         = {Machine review of arXiv:2508.09909}
}
read the original abstract

This SHREC 2025 track focuses on the recognition and segmentation of relief patterns embedded on the surface of a set of synthetically generated triangle meshes. We report the methods proposed by the participants, whose performance highlights the inherent complexity of solving the problem, which is still open. Then, we discuss the critical aspects of the proposed tasks, highlight the limitations of current techniques, and outline possible directions for future research. All resources and track details are available at the official track webpage: https://sites.google.com/unifi.it/shrec25-relief-pattern.

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

47 extracted references · 36 canonical work pages

  1. [1]

    Convolutional neural network for pottery retrieval

    Halim Benhabiles and Hedi Tabia. “Convolutional neural network for pottery retrieval”. In: Journal of Electronic Imaging 26.1 (2016), p. 011005. DOI: 10 . 1117 / 1 . JEI . 26 . 1 . 011005 . URL: https://doi.org/10.1117/1.JEI.26.1. 011005

    work page doi:10.1117/1.jei.26.1 2016

  2. [2]

    SHREC’17 Track: Retrieval of Surfaces With Similar Relief Patterns

    Silvia Biasotti et al. “SHREC’17 Track: Retrieval of Surfaces With Similar Relief Patterns”. In: Apr. 2017

    work page 2017

  3. [3]

    SHREC’18 track: Recognition of geometric patterns over 3D models

    Silvia Biasotti et al. “SHREC’18 track: Recognition of geometric patterns over 3D models”. In: Euro- graphics workshop on 3D object retrieval. 2018

    work page 2018

  4. [4]

    Blender - a 3D mod- elling and rendering package

    Blender Online Community. Blender - a 3D mod- elling and rendering package . Accessed October

    work page

  5. [5]

    Instance segmentation of 3D woven fabric from tomography images by Deep Learning and morphological pseudo-labeling

    Samy Blusseau et al. “Instance segmentation of 3D woven fabric from tomography images by Deep Learning and morphological pseudo-labeling”. In: Composites Part B: Engineering 247 (2022), p. 110333. ISSN : 1359-8368. DOI: https : / / doi . org / 10 . 1016 / j . compositesb . 2022 . 110333 . URL: https : / / www . sciencedirect.com/science/article/ pii/S1359...

    work page 2022

  6. [6]

    Geometric Deep Learn- ing: Going beyond Euclidean data

    Michael M. Bronstein et al. “Geometric Deep Learn- ing: Going beyond Euclidean data”. In: IEEE Signal Processing Magazine 34.4 (2017), pp. 18–42. DOI: 10.1109/MSP.2017.2693418

    work page arXiv 2017

  7. [7]

    Deep Learning-Based Engrav- ing Segmentation of 3-D Inscriptions Extracted From the Rough Surface of Ancient Stelae

    Ye-Chan Choi et al. “Deep Learning-Based Engrav- ing Segmentation of 3-D Inscriptions Extracted From the Rough Surface of Ancient Stelae”. In: IEEE Ac- cess 9 (2021), pp. 153199–153212. DOI: 10.1109/ ACCESS.2021.3127229. 10

    work page arXiv 2021

  8. [8]

    Objaverse-XL: A Uni- verse of 10M+ 3D Objects

    Matt Deitke et al. “Objaverse-XL: A Uni- verse of 10M+ 3D Objects”. In: arXiv preprint arXiv:2307.05663 (2023)

    arXiv 2023

Show all 47 references
  1. [9]

    Three-dimensional solid texture analysis in biomedical imaging: Review and opportunities

    Adrien Depeursinge et al. “Three-dimensional solid texture analysis in biomedical imaging: Review and opportunities”. In: Medical Image Analysis 18.1 (2014), pp. 176–196. ISSN : 1361-8415. DOI: https : / / doi . org / 10 . 1016 / j . media . 2013 . 10 . 005 . URL: https : / / ...

    2014

  2. [10]

    Probing the 3d awareness of visual foundation models

    Mohamed El Banani et al. “Probing the 3d awareness of visual foundation models”. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pat- tern Recognition. 2024, pp. 21795–21806

    2024

  3. [11]

    Labeled Facets: New Surface Texture Dataset

    Iyyakutti Iyappan Ganapathi and Naoufel Werghi. “Labeled Facets: New Surface Texture Dataset”. In: 2022. URL: https : / / api . semanticscholar . org / CorpusID : 254428249

    2022

  4. [12]

    3D-TexSeg: Un- supervised Segmentation of 3D Texture using Mutual Transformer Learning

    Iyyakutti Iyappan Ganapathi et al. 3D-TexSeg: Un- supervised Segmentation of 3D Texture using Mutual Transformer Learning. 2023. arXiv: 2311.10651 [cs.CV]. URL: https://arxiv.org/abs/ 2311.10651

    2023 arXiv

  5. [13]

    Detecting 3D Texture on Cultural Heritage Artifacts

    Iyyakutti Iyappan Ganapathi et al. “Detecting 3D Texture on Cultural Heritage Artifacts”. In: Pat- tern Recognition, Computer Vision, and Image Pro- cessing. ICPR 2022 International Workshops and Challenges. Ed. by Jean-Jacques Rousseau and Bill Kapralos. Cham: Springer Nature...

    2022

  6. [14]

    Graph Based Texture Pattern Classification

    Iyyakutti Iyappan Ganapathi et al. “Graph Based Texture Pattern Classification”. In:2022 8th Interna- tional Conference on Virtual Reality (ICVR) . 2022, pp. 363–369. DOI: 10 . 1109 / ICVR55215 . 2022.9847889

    2022

  7. [15]

    Unsupervised Dual Transformer Learning for 3-D Textured Sur- face Segmentation

    Iyyakutti Iyappan Ganapathi et al. “Unsupervised Dual Transformer Learning for 3-D Textured Sur- face Segmentation”. In: IEEE Transactions on Neu- ral Networks and Learning Systems 36.3 (2025), pp. 5020–5031. DOI: 10 . 1109 / TNNLS . 2024 . 3365515

    2025

  8. [16]

    Texture Analysis and Its Applications in Biomedical Imaging: A Sur- vey

    Maryam Khaksar Ghalati et al. “Texture Analysis and Its Applications in Biomedical Imaging: A Sur- vey”. In: IEEE Reviews in Biomedical Engineering 15 (2022), pp. 222–246. DOI: 10 . 1109 / RBME . 2021.3115703

    2022

  9. [17]

    Effective characterization of relief patterns

    Andrea Giachetti. “Effective characterization of relief patterns”. In: Computer Graphics Forum . V ol. 37. 5. Wiley Online Library. 2018, pp. 83–92

    2018

  10. [18]

    Deep Residual Learning for Im- age Recognition

    Kaiming He et al. “Deep Residual Learning for Im- age Recognition”. In: 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) . 2016, pp. 770–778. DOI: 10.1109/CVPR.2016. 90

    2016 doi

  11. [19]

    Convolutional Networks with Dense Connectivity

    Gao Huang et al. “Convolutional Networks with Dense Connectivity”. In: IEEE Transactions on Pat- tern Analysis and Machine Intelligence44.12 (2022), pp. 8704–8716. DOI: 10 . 1109 / TPAMI . 2019 . 2918284

    2022

  12. [20]

    Near-Regular Structure Discovery Using Lin- ear Programming

    Qixing Huang, Leonidas J. Guibas, and Niloy J. Mi- tra. “Near-Regular Structure Discovery Using Lin- ear Programming”. In: ACM Trans. Graph. 33.3 (June 2014). ISSN : 0730-0301. DOI: 10 . 1145 / 2535596. URL: https : / / doi . org / 10 . 1145/2535596

    2014

  13. [21]

    The Efficacy Analy- sis of Determining the Wooded and Shrubbed Area Based on Archival Aerial Imagery Using Texture Analysis

    Przemysław Kupidura et al. “The Efficacy Analy- sis of Determining the Wooded and Shrubbed Area Based on Archival Aerial Imagery Using Texture Analysis”. In: ISPRS International Journal of Geo- Information 8.10 (2019). ISSN : 2220-9964. DOI: 10. 3390 / ijgi8100450. URL: https ...

    2019

  14. [22]

    Visualization and Extraction of Carvings for Heritage Conservation

    Kai Lawonn et al. “Visualization and Extraction of Carvings for Heritage Conservation”. In: IEEE Transactions on Visualization and Computer Graph- ics 23.1 (2017), pp. 801–810. DOI: 10 . 1109 / TVCG.2016.2598603

    2017

  15. [23]

    OpenShape: Scaling Up 3D Shape Representation Towards Open-World Under- standing

    Minghua Liu et al. OpenShape: Scaling Up 3D Shape Representation Towards Open-World Under- standing. 2023. arXiv: 2305.10764 [cs.CV]

    2023 arXiv

  16. [24]

    Segmenting Geometric Reliefs from Textured Background Surfaces

    Sheng-Dong Liu et al. “Segmenting Geometric Reliefs from Textured Background Surfaces”. In: Computer-aided Design and Applications 4 (2007), pp. 565–583. URL: https : / / api . semanticscholar . org / CorpusID : 6248423

    2007

  17. [25]

    Segmenting periodic reliefs on triangle meshes

    Shenglan Liu et al. “Segmenting periodic reliefs on triangle meshes”. In:Proceedings of the 12th IMA In- ternational Conference on Mathematics of Surfaces XII. Sheffield, UK: Springer-Verlag, 2007, pp. 290–

    2007

  18. [26]

    Segmenting reliefs on trian- gle meshes

    Shenglan Liu et al. “Segmenting reliefs on trian- gle meshes”. In: Proceedings of the 2006 ACM Symposium on Solid and Physical Modeling . SPM ’06. Cardiff, Wales, United Kingdom: Associa- tion for Computing Machinery, 2006, pp. 7–16. ISBN : 1595933581. DOI: 10 . 1145 / 1128888...

    2006

  19. [27]

    Texture Transfer Using Ge- ometry Correlation

    Tom Mertens et al. “Texture Transfer Using Ge- ometry Correlation”. In: Symposium on Rendering . Ed. by Tomas Akenine-Moeller and Wolfgang Hei- drich. The Eurographics Association, 2006. ISBN : 3- 905673-35-5. DOI: /10.2312/EGWR/EGSR06/ 273-284

    2006 doi

  20. [28]

    https : / / people

    MIT CSAIL Textured Models Database . https : / / people . csail . mit . edu / tmertens / textransfer / data/ . Accessed April 2025. 2008

    2025

  21. [29]

    SHREC 2020: Retrieval of digital surfaces with similar geomet- ric reliefs

    Elia Moscoso Thompson et al. “SHREC 2020: Retrieval of digital surfaces with similar geomet- ric reliefs”. In: Computers & Graphics 91 (2020), pp. 199–218. ISSN : 0097-8493. DOI: https : / / doi . org / 10 . 1016 / j . cag . 2020 . 07 . 011 . URL: https : / / www . sciencedire...

    2020

  22. [30]

    DINOv2: Learning Robust Visual Features without Supervision

    Maxime Oquab et al. DINOv2: Learning Robust Visual Features without Supervision . 2024. arXiv: 2304 . 07193 [cs.CV] . URL: https : / / arxiv.org/abs/2304.07193

    2024 arXiv

  23. [31]

    Single tree species classi- fication from Terrestrial Laser Scanning data for forest inventory

    Ahlem Othmani et al. “Single tree species classi- fication from Terrestrial Laser Scanning data for forest inventory”. In: Pattern Recognition Letters 34.16 (2013), pp. 2144–2150. ISSN : 0167-8655. DOI: https : / / doi . org / 10 . 1016 / j . patrec.2013.08.004 . URL: https://...

    2013

  24. [32]

    Binary segmentation of relief patterns on point clouds

    Gabriele Paolini, Claudio Tortorici, and Stefano Berretti. “Binary segmentation of relief patterns on point clouds”. In: Computers & Graphics 123 (2024), p. 104020. ISSN : 0097-8493. DOI: https: / / doi . org / 10 . 1016 / j . cag . 2024 . 104020 . URL: https : / / www . scien...

    2024

  25. [33]

    CC0 PBR Textures

    Joao Paulo. CC0 PBR Textures. https : / / www . patreon.com/gendo . Materal library available under CC0 License via Patreon, Accessed October

  26. [34]

    Improving the Fisher Kernel for Large- Scale Image Classification

    Florent Perronnin, Jorge S ´anchez, and Thomas Mensink. “Improving the Fisher Kernel for Large- Scale Image Classification”. In: Computer Vision – ECCV 2010 . Ed. by Kostas Daniilidis, Petros Maragos, and Nikos Paragios. Berlin, Heidelberg: Springer Berlin Heidelberg, 2010, pp...

    2010

  27. [35]

    https : / / polyhaven

    Polyhaven. https : / / polyhaven . com . Ac- cessed November 2024. n.d

    2024

  28. [36]

    Learning Transferable Visual Models From Natural Language Supervision

    Alec Radford et al. Learning Transferable Visual Models From Natural Language Supervision . 2021. arXiv: 2103.00020 [cs.CV] . URL: https:// arxiv.org/abs/2103.00020

    2021 arXiv

  29. [37]

    DiffusionNet: Discretization Agnostic Learning on Surfaces

    Nicholas Sharp et al. DiffusionNet: Discretization Agnostic Learning on Surfaces. 2022. arXiv: 2012. 00888 [cs.CV]. URL: https://arxiv.org/ abs/2012.00888

    2022 arXiv

  30. [38]

    Very Deep Convolutional Networks for Large-Scale Image Recognition

    Karen Simonyan and Andrew Zisserman. Very Deep Convolutional Networks for Large-Scale Image Recognition. 2015. arXiv: 1409.1556 [cs.CV] . URL: https : / / arxiv . org / abs / 1409 . 1556

    2015 arXiv

  31. [39]

    https : / / sketchfab

    Sketchfab. https : / / sketchfab . com . Ac- cessed November 2024. n.d

    2024

  32. [40]

    Mingxing Tan and Quoc V . Le. EfficientNet: Rethink- ing Model Scaling for Convolutional Neural Net- works. 2020. arXiv: 1905.11946 [cs.LG]. URL: https://arxiv.org/abs/1905.11946

    2020 arXiv

  33. [41]

    Retrieval of gray patterns depicted on 3D models

    E. Moscoso Thompson et al. “Retrieval of gray patterns depicted on 3D models”. In: Proceedings of the 11th Eurographics Workshop on 3D Ob- ject Retrieval . 3DOR ’18. Delft, The Netherlands: Eurographics Association, 2018, pp. 63–69. ISBN : 9783038680536

    2018

  34. [42]

    The Mesh-LBP: A Framework for Extract- ing Local Binary Patterns From Discrete Manifolds

    Naoufel Werghi, Stefano Berretti, and Alberto del Bimbo. “The Mesh-LBP: A Framework for Extract- ing Local Binary Patterns From Discrete Manifolds”. In: IEEE Transactions on Image Processing 24.1 (2015), pp. 220–235. DOI: 10.1109/TIP.2014. 2370253

    2015 doi

  35. [43]

    ULIP-2: Towards Scalable Multi- modal Pre-training for 3D Understanding

    Le Xue et al. ULIP-2: Towards Scalable Multi- modal Pre-training for 3D Understanding . 2024. arXiv: 2305.08275 [cs.CV] . URL: https:// arxiv.org/abs/2305.08275

    2024 arXiv

  36. [44]

    Guest Editorial: Non- Euclidean Machine Learning

    Stefanos Zafeiriou et al. “Guest Editorial: Non- Euclidean Machine Learning”. In: IEEE Transac- tions on Pattern Analysis and Machine Intelligence 44.2 (2022), pp. 723–726. DOI: 10.1109/TPAMI. 2021.3129857

    2022

  37. [45]

    iBOT: Image BERT Pre-Training with Online Tokenizer

    Jinghao Zhou et al. iBOT: Image BERT Pre-Training with Online Tokenizer. 2022. arXiv: 2111.07832 [cs.CV]. URL: https://arxiv.org/abs/ 2111.07832. 12

    2022 arXiv

  38. [306]

    ISBN : 9783540738428

  39. [2024]

    Blender Institute, Ams- terdam, n.d

    Blender Foundation. Blender Institute, Ams- terdam, n.d

Pith tools

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