Pith. sign in

REVIEW 5 major objections 5 minor 37 references

Recognizing Artistic Style of Archaeological Image Fragments Using Deep Style Extrapolation

T0 review · 5 major / 5 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 A style extrapolation preprocessing step lets fine-tuned deep networks classify artistic style of image fragments, beating prior methods on CLEOPATRA and a new Pompeii benchmark.

desk verdict Genuinely useful idea and dataset, but the missing POMPAAF split protocol is a load-bearing hole in the experimental claims. read the letter →

arxiv 2501.00836 v2 pith:V6HQE2VY submitted 2025-01-01 cs.CV

Gur Elkin , Ofir Itzhak Shahar , Yaniv Ohayon , Nadav Alali , Ohad Ben-Shahar This is my paper · ORCID
classification cs.CV
keywords imageclassificationartisticstylerecognitionfrescofragmentstransferdeeplearningPOMPAAFdatasetculturalheritageEfficientNet
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

The paper argues that a two-stage deep learning pipeline can identify the artistic style of small broken image fragments, a task that stumps non-expert humans and prior machine learning methods. The first stage extrapolates the fragment's style outward to fill the whole image frame with a stylistically consistent synthetic picture, and the second stage classifies that filled image with a fine-tuned EfficientNet. On the CLEOPATRA benchmark, the method reaches 0.475 accuracy versus 0.406 for transfer learning alone and 0.28 for the prior random forest; on the new POMPAAF dataset, it reaches up to 0.967 accuracy on square fragments. If correct, the approach would make automated re-sorting of mixed fresco fragments from archaeological sites more reliable.

What carries the argument

The style extrapolation module is the load-bearing component. It is a modified autoencoder with a pre-trained ResNet18 encoder, a convolutional decoder, and a residual connection, optimized by the loss $L_{\mathrm{SE}} = \lambda L_{\mathrm{AS}} + \mu L_{\mathrm{MC}}$. Here $L_{\mathrm{AS}}$ is the auto-style loss comparing Gram matrices of VGG activations of the input and output, and $L_{\mathrm{MC}}$ is a masked content loss that forces the output to stay close to the input on fragment pixels while allowing the background to be filled. This turns an irregular fragment with an arbitrary background into a stylistically consistent full-frame image, which the classifier can then process without shape confounding.

What would settle it

Inspect the POMPAAF split: if the train and test sets share fragments cut from the same source fresco, retrain the pipeline with a fresco-disjoint split and compare accuracies; a large drop would indicate the model memorized individual frescos instead of learning style, while holding out entire walls and testing whether accuracy stays above chance would settle the question directly.

Watch

Extended reading notes

Core claim

The central claim is that a style extrapolation module—a convolutional autoencoder trained with a Gram-matrix style loss and a masked content loss—can diffuse a fragment's stylistic features into the non-pictorial background, so a subsequent classifier sees a complete image rather than an arbitrary silhouette. The paper reports state-of-the-art results with this module feeding a fine-tuned EfficientNet: 0.475 accuracy on CLEOPATRA's 11 styles and between 0.846 and 0.967 accuracy across the four POMPAAF fragmentation geometries, always beating both the simple CNN baseline and transfer learning alone. Smaller and more numerous fragments consistently yield higher accuracy than larger pieces, which the authors attribute to reduced downscaling at a fixed input resolution. The paper also introduces POMPAAF, a dataset of 311 Pompeian fresco images artificially broken into four shape families.

Load-bearing premise

The reported POMPAAF accuracies assume that no fragments from the same source fresco appear in both the training and test sets, since the paper does not describe the split protocol and the model could otherwise memorize each fresco's distinctive appearance rather than learn a general artistic style.

Editorial extensions

If this is right

  • On the CLEOPATRA dataset, the full pipeline's 0.475 accuracy beats transfer learning alone (0.406) and the prior random forest (0.28), indicating that the style extrapolation step is the source of the gain.
  • On POMPAAF, the method exceeds 0.94 accuracy for 160-piece square and non-convex fragment sets, suggesting fragment shape is not a barrier once style is extrapolated.
  • More fragments per fresco helps accuracy: for square fragments, accuracy rises from 0.846 at 12 pieces to 0.967 at 160 pieces, so smaller pieces do not lose information at a fixed input resolution.
  • Across crossing-cuts, non-convex, and eroded Voronoi fragment geometries, the proposed method outperforms both baselines by substantial margins, indicating the approach generalizes beyond square pieces.
  • The introduced POMPAAF dataset provides a new controlled benchmark for fragment style classification with varied fragment geometry.

Reading between the lines

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

  • Editorial inference: the paper never describes the POMPAAF train/test split protocol; if fragments from the same fresco appear in both sets, the classifier could memorize per-fresco colors and textures, inflating the reported accuracies.
  • Editorial inference: a fresco-disjoint split—training on some walls and testing on fragments from unseen walls—would directly test whether the model learns a general Pompeian style rather than instance-specific appearance.
  • Editorial inference: coupling the style extrapolation idea with diffusion-based inpainting or attention-based classifiers, which the paper names as future directions, could push accuracy higher or reduce data requirements, though this is not tested here.
  • Editorial inference: because CLEOPATRA's classes are historical eras, the same pipeline might transfer to other fragment domains such as pottery or manuscript fragments, but the paper provides no evidence for that transfer.
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

5 major / 5 minor

Summary. The paper proposes a two-stage pipeline for classifying the artistic style of image fragments: a "style extrapolation" preprocessing module, implemented as a ResNet18-based autoencoder with a residual connection and trained with a self-referential style loss (Eq. 3) plus a masked content loss (Eq. 4), followed by a fine-tuned EfficientNet classifier. The method is evaluated on the existing CLEOPATRA benchmark and on POMPAAF, a new dataset synthesized by fragmenting 311 Pompeian fresco images with four fragmentation methods at several piece counts. The authors report accuracy of 0.475 on the CLEOPATRA test set versus 0.406 for transfer learning alone and 0.28 for the prior random-forest model, and up to 0.967 accuracy on the square-fragment POMPAAF subset. The central claim is that the style extrapolation module improves fragment style classification over transfer learning and prior art.

Significance. If the reported results are reproducible, the work offers a useful preprocessing idea and a new public benchmark for a niche but practically relevant task. The POMPAAF dataset and code release are concrete contributions that could support future comparisons. The main claims, however, are entirely empirical and currently rest on evaluation details that are not fully documented. The paper contains no theoretical or machine-checkable results; its value hinges on whether the reported accuracy gains are genuine and not an artifact of data leakage or of inadequate baselines.

major comments (5)
  1. [Section 5.3, Tables 2-5] The paper never states how POMPAAF was partitioned into training and test sets. Since the dataset is generated by fragmenting 311 source fresco images, a per-fragment random split can place fragments from the same wall in both train and test. A classifier could then memorize per-fresco color, texture, and condition cues and inflate accuracy without learning a generalizable notion of style. This is load-bearing because all POMPAAF results are the main evidence for the claimed advantage on 'varying styles and geometries.' Please specify whether the split is by source fresco image or by individual fragment, report test-set sizes and class balance, and if the current split is per-fragment, re-run the experiments with a source-disjoint split.
  2. [Section 5.2 and 5.3] No experiment is repeated or reported with variance. The CLEOPATRA test set has only 80 pieces, so the 0.475 versus 0.406 gap may be within random noise, and the POMPAAF tables report single runs without error bars. Please provide confidence intervals, repeated-seed results, or statistical significance tests, and report per-class metrics for POMPAAF, especially for the small-piece regimes.
  3. [Section 4.1, Eq. (5)] The weights lambda and mu in the combined style extrapolation loss LSE = lambda*LAS + mu*LMC are hand-chosen, but no values are given, and no sensitivity analysis is reported. Together with the missing training set, epoch count, optimizer, and learning rate for the extrapolator, this makes the method not reproducible as written. Please add these details.
  4. [Section 4.1, Eqs. (3)-(5)] Because Eq. (3) forces the output style to match the input style and the residual connection sums the input and the network output, an identity mapping is a trivial solution that satisfies both losses (the content loss is masked to fragment pixels). The paper provides no example outputs or quantitative evidence that the extrapolator actually fills background regions rather than degenerating. Please report qualitative visualizations of extrapolated fragments and a quantitative analysis of how much the background is modified.
  5. [Section 4.3 and 5.3] The transfer-learning baseline is compared against the proposed method, but it is unclear whether the baseline receives the same input size, background handling, and preprocessing as the proposed method. Without an ablation that replaces the extrapolator with simple padding, cropping, or inpainting to the classifier's input size, the observed accuracy gains cannot be attributed to the extrapolation module rather than to a more favorable input representation.
minor comments (5)
  1. [Throughout] There are numerous typos, including 'Forbenius' in Eq. (1), 'frgments' in Fig. 2, 'corssing-cuts' in Section 5.3, 'Pomepian' in Section 3, and 'CLEOPARTA' in Section 6. Please proofread carefully.
  2. [Table 4] The 12-piece Precision row is written as '0.45' while all other entries use three decimals; please format all table entries consistently.
  3. [Section 5.1, Eq. (7)] Eq. (7) defines accuracy through a binary confusion matrix, but the evaluation is multi-class. Please clarify whether the reported accuracy is overall accuracy or per-class accuracy and whether any macro-averaging is applied to accuracy itself.
  4. [Section 5.2, Table 1] The comparison with Cascone et al. relies on numbers from the original paper. Please state explicitly whether the same CLEOPATRA train/test split and input preprocessing were used, and whether the random-forest baseline was re-run or only cited.
  5. [Table 1 footnote] The footnote says recall was calculated from precision and F1. For macro-averaged metrics this inference is only valid if the same averaging was used; please confirm the calculation.

Circularity Check

0 steps flagged · score 1.0 of 10

No circular derivation: classification accuracy is measured on held-out fragments, and the style extrapolator's self-referential loss is a preprocessing objective, not a label prediction.

full rationale

The paper's derivation chain is: train a style extrapolator with hand-crafted losses (Eqs. 3-5), freeze it, fine-tune a pre-trained EfficientNet classifier with cross-entropy (Eq. 6), and measure accuracy on held-out test sets. The auto-style loss (Eq. 3) compares Gram matrices of the input I and the extrapolator output E(I), so it enforces style preservation rather than predicting a style label. The masked content loss (Eq. 4) likewise preserves fragment pixels. These losses are not fitted to classification labels, and the extrapolator does not directly emit the classification output. The central claim is the downstream classification accuracy on CLEOPATRA and POMPAAF test sets, which is an empirical measurement, not a quantity that reduces to the training objective by construction. The comparison against transfer learning without the extrapolator is a controlled ablation, and the comparison against the prior random forest result on CLEOPATRA uses an external benchmark. The paper's self-citations (e.g., Harel et al. for crossing-cuts fragmentation, Napolitano et al. as the Pompeii image source, Tsesmelis et al. for RePAIR context) provide data or generative algorithms and are not used to justify the central classification claim. No uniqueness theorem or ansatz is imported from same-author work. The underspecified POMPAAF train/test split is a legitimate correctness and leakage concern, but the paper does not state the split protocol, so one cannot exhibit a specific reduction of the reported accuracies to training-set memorization from the text; this is therefore a validity risk rather than a demonstrated circular step.

Assumptions & free parameters 2 free parameters · 4 assumptions · 0 invented entities

The central approach combines existing style-transfer losses and transfer learning; it introduces no new physical entities. The main unverified inputs are the hand-chosen loss weights and the validity of Pompeian style labels and synthetic fragmentation as proxies for real fragments. The reliance on the authors' own earlier dataset and fragmentation code adds a mild self-citation burden but does not make the classification result circular, since the final accuracy is measured on a test set.

free parameters (2)
  • lambda (style loss weight in eq 5) = not reported
    L_SE = lambda * L_AS + mu * L_MC; the balance between style preservation and content preservation affects extrapolation quality and downstream accuracy, but no value or tuning procedure is given.
  • mu (masked content loss weight in eq 5) = not reported
    Weights the masked content loss that keeps original fragment pixels unchanged; no value or tuning procedure is given.
assumptions (4)
  • domain assumption The August Mau four-style taxonomy of Pompeian wall paintings is a valid and discriminative label set.
    The paper labels fragments by these four styles, relying on art-historical categories rather than independently verifying that the styles are visually separable in fragment images.
  • domain assumption The synthetic fragmentation methods (square, crossing cuts, non-convex, eroded Voronoi) produce fragments representative of real archaeological fragments.
    The POMPAAF dataset is generated with these algorithms; if the fragments differ from real ones, reported accuracy may not transfer to real archaeological scenarios.
  • domain assumption ImageNet-pretrained features transfer to ancient Roman fresco style classification.
    The encoder (ResNet18) and classifier (EfficientNet) are pre-trained on ImageNet, and the paper assumes low-level features are shared between natural images and fresco fragments.
  • domain assumption The alpha channel provides an appropriate mask M for separating fragment pixels from background.
    Eq (4) uses M to weight content preservation; if the mask is inaccurate, the extrapolator may alter fragment content.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Recognizing Artistic Style of Archaeological Image Fragments Using Deep Style Extrapolation." pith.science (2026). https://pith.science/paper/V6HQE2VY

@misc{pith2026250100836,
  author       = {Pith},
  title        = {Pith review of: Recognizing Artistic Style of Archaeological Image Fragments Using Deep Style Extrapolation},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/V6HQE2VY}},
  note         = {Machine review of arXiv:2501.00836}
}
read the original abstract

Ancient artworks obtained in archaeological excavations usually suffer from a certain degree of fragmentation and physical degradation. Often, fragments of multiple artifacts from different periods or artistic styles could be found on the same site. With each fragment containing only partial information about its source, and pieces from different objects being mixed, categorizing broken artifacts based on their visual cues could be a challenging task, even for professionals. As classification is a common function of many machine learning models, the power of modern architectures can be harnessed for efficient and accurate fragment classification. In this work, we present a generalized deep-learning framework for predicting the artistic style of image fragments, achieving state-of-the-art results for pieces with varying styles and geometries.

Figures

Figures reproduced from arXiv: 2501.00836 by the authors.

Figure 1
Figure 1. Frescos from Pompeii, representing the four styles of ancient Roman wall paintings, sampled from the Pompeii Archives published by Napolitano et al. [24]. 3.2 POMPAAF Dataset Description Not only do ancient Roman frescos provide us with a rare peek into the civiliza￾tion and culture of roughly 2,000 years ago, but their four distinct artistic styles make an interesting test case for the art style classification task… view at source ↗
Figure 2
Figure 2. Examples of the four fragmentation methods used to create POMPAAF. The methods are demonstrated on a fresco of the 4th Pompeian style, originally located in House of the Vettii, Pompeii. (a) Strictly square fragments. (b) Crossing cuts polygonal fragments. (c) Non-convex partition fragments. (d) Eroded Voronoi partition frgments [PITH_FULL_IMAGE:figures/full_fig_p006_2.png] view at source ↗
Figure 3
Figure 3. Diagram of the proposed model. Yellow (top-left): Style extrapolation mod￾ule. Green (top-right): Pre-trained image classifier. Purple (bottom): Loss com￾putation for the style extrapolator. The depicted fragment is from the CLEOPATRA [6] training set [PITH_FULL_IMAGE:figures/full_fig_p007_3.png] view at source ↗

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

37 extracted references · 31 canonical work pages

  1. [1]

    https://smarthistory.org/ roman-wall-painting-styles/ (Jun 2018), accessed: 2024-06-17 4

    Ambler, D.J.L.: Roman wall painting styles. https://smarthistory.org/ roman-wall-painting-styles/ (Jun 2018), accessed: 2024-06-17 4

    work page 2018

  2. [2]

    IEEE PAMI39(2), 385–396 (2016) 6

    Andalo, F., Taubin, G., Goldenstein, S.: PSQP: Puzzle solving by quadratic pro- gramming. IEEE PAMI39(2), 385–396 (2016) 6

    work page 2016

  3. [3]

    Thames & Hudson, London (2013) 4

    Berry, J.: The Complete Pompeii. Thames & Hudson, London (2013) 4

    work page 2013

  4. [4]

    Machine learning45, 5–32 (2001) 3, 10

    Breiman, L.: Random forests. Machine learning45, 5–32 (2001) 3, 10

    work page 2001

  5. [5]

    In: International Conference on Image Analysis and Processing

    Cascone, L., Dondi, P., Lombardi, L., Narducci, F.: Automatic classification of fresco fragments: a machine and deep learning study. In: International Conference on Image Analysis and Processing. pp. 701–712. Springer (2022) 2, 3

    work page 2022

  6. [6]

    Journal of Ambient Intelligence and Humanized Com- puting 14(4), 4087–4097 (2023) 2, 3, 4, 7, 9, 10, 11, 12

    Cascone, L., Nappi, M., Narducci, F., Russo, S.L.: Classification of fragments: recognition of artistic style. Journal of Ambient Intelligence and Humanized Com- puting 14(4), 4087–4097 (2023) 2, 3, 4, 7, 9, 10, 11, 12

    work page 2023

  7. [7]

    In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recog- nition

    Chawla, P., Jandial, S., Badjatiya, P., Chopra, A., Sarkar, M., Krishnamurthy, B.: Leveraging style and content features for text conditioned image retrieval. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recog- nition. pp. 3978–3982 (2021) 1

    work page 2021

  8. [8]

    250: Ritual, Space, and Decoration

    Clarke, J.R.: The Houses of Roman Italy, 100 B.C.-A.D. 250: Ritual, Space, and Decoration. University of California Press, Los Angeles (1991), explores how the preservation of Pompeiian art provides insights into Roman culture and civilization 4

    work page 1991

Show all 37 references
  1. [9]

    In: 2009 IEEE conference on computer vision and pattern recognition

    Deng, J., Dong, W., Socher, R., Li, L.J., Li, K., Fei-Fei, L.: Imagenet: A large- scale hierarchical image database. In: 2009 IEEE conference on computer vision and pattern recognition. pp. 248–255. Ieee (2009) 9

    2009

  2. [10]

    Pattern Recognition Letters138, 631–637 (2020) 3, 6

    Dondi, P., Lombardi, L., Setti, A.: Dafne: A dataset of fresco fragments for digital anastlylosis. Pattern Recognition Letters138, 631–637 (2020) 3, 6

    2020

  3. [11]

    arXiv preprint arXiv:2010.11929 (2020) 12

    Dosovitskiy, A., Beyer, L., Kolesnikov, A., Weissenborn, D., Zhai, X., Unterthiner, T., Dehghani, M., Minderer, M., Heigold, G., Gelly, S., et al.: An image is worth 16x16 words: Transformers for image recognition at scale. arXiv preprint arXiv:2010.11929 (2020) 12

    2020 arXiv

  4. [12]

    In: Proceedings of the IEEE conference on computer vision and pattern recognition

    Gatys, L.A., Ecker, A.S., Bethge, M.: Image style transfer using convolutional neural networks. In: Proceedings of the IEEE conference on computer vision and pattern recognition. pp. 2414–2423 (2016) 3, 8

    2016

  5. [13]

    International Journal of Computer Vision (2024)

    Harel, P., Shahar, O.I., Ben-Shahar, O.: Pictorial and apictorial polygonal jig- saw puzzles from arbitrary number of crossing cuts. International Journal of Computer Vision (2024). https://doi.org/10.1007/s11263-024-02033-7, https:// doi.org/10.1007/s11263-024-02033-7 6

    2024 doi

  6. [14]

    He,K.,Zhang,X.,Ren,S.,Sun,J.:Deepresiduallearningforimagerecognition.In: Proceedings of the IEEE conference on computer vision and pattern recognition. pp. 770–778 (2016) 2, 8 16 G. Elkin et al

    2016

  7. [15]

    arXiv preprint arXiv:1311.3715 (2013) 3

    Karayev, S., Trentacoste, M., Han, H., Agarwala, A., Darrell, T., Hertzmann, A., Winnemoeller, H.: Recognizing image style. arXiv preprint arXiv:1311.3715 (2013) 3

    2013 arXiv

  8. [16]

    In: Asian conference on machine learning

    Lecoutre, A., Negrevergne, B., Yger, F.: Recognizing art style automatically in painting with deep learning. In: Asian conference on machine learning. pp. 327–

  9. [17]

    arXiv preprint arXiv:1701.01036 (2017) 12

    Li, Y., Wang, N., Liu, J., Hou, X.: Demystifying neural style transfer. arXiv preprint arXiv:1701.01036 (2017) 12

    2017 arXiv

  10. [18]

    Cambridge University Press, New York (1991) 4

    Ling, R.: Roman Painting. Cambridge University Press, New York (1991) 4

    1991

  11. [19]

    Liu, Z., Lin, Y., Cao, Y., Hu, H., Wei, Y., Zhang, Z., Lin, S., Guo, B.: Swin transformer:Hierarchical visiontransformerusingshiftedwindows.In:Proceedings of the IEEE/CVF international conference on computer vision. pp. 10012–10022 (2021) 12

    2021

  12. [20]

    The American Journal of Philology 126(4 (Whole Number 504)), 626–629 (2005)

    Lorenz, K.G.: The social life of painting in ancient rome and on the bay of naples (review). The American Journal of Philology 126(4 (Whole Number 504)), 626–629 (2005). https://doi.org/10.1353/ajp.2006. 0010, https://www.researchgate.net/publication/236824125_The_Social_Life_...

    2005

  13. [21]

    In: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition

    Lugmayr, A., Danelljan, M., Romero, A., Yu, F., Timofte, R., Van Gool, L.: Re- paint: Inpainting using denoising diffusion probabilistic models. In: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition. pp. 11461– 11471 (2022) 12

    2022

  14. [22]

    Reimer, Berlin (1882) 4

    Mau, A.: Geschichte der decorativen Wandmalerei in Pompeji. Reimer, Berlin (1882) 4

  15. [23]

    The MacMillan Company, New York (1902) 4

    Mau, A.: Pompeii: Its Life and Art. The MacMillan Company, New York (1902) 4

    1902

  16. [24]

    In: Proceedings of Pictura Parietum

    Napolitano, M., Giudice, S., Gravina, E., Scarpati, G., Zambrano, A., Zuchtriegel, G., Aslan, S., Khoroshiltseva, M., Palmieri, L., Pelillo, M., Vascon, S., Ben-Shahar, O., Ohayon, Y., Shahar, O., Traviglia, A.: The dataset of paintings from the pom- peii archive: A task of th...

    2023

  17. [25]

    2015 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) pp

    Paikin, G., Tal, A.: Solving multiple square jigsaw puzzles with missing pieces. 2015 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) pp. 4832–4839 (2015) 6

    2015

  18. [26]

    In: CVPR 2011

    Pomeranz, D., Shemesh, M., Ben-Shahar, O.: A fully automated greedy square jigsaw puzzle solver. In: CVPR 2011. pp. 9–16. IEEE (2011) 6

    2011

  19. [27]

    Journal of King Saud University-Computer and Infor- mation Sciences 32(8), 883–894 (2020) 3

    Rasheed, N.A., Nordin, M.J.: Classification and reconstruction algorithms for the archaeological fragments. Journal of King Saud University-Computer and Infor- mation Sciences 32(8), 883–894 (2020) 3

    2020

  20. [28]

    arXiv preprint arXiv:1409.1556 (2014) 2, 8

    Simonyan, K., Zisserman, A.: Very deep convolutional networks for large-scale image recognition. arXiv preprint arXiv:1409.1556 (2014) 2, 8

    2014 arXiv

  21. [29]

    In: 2010 IEEE Computer Society Conference on Computer Vision and Pattern Recognition- Workshops

    Smith, P., Bespalov, D., Shokoufandeh, A., Jeppson, P.: Classification of archae- ological ceramic fragments using texture and color descriptors. In: 2010 IEEE Computer Society Conference on Computer Vision and Pattern Recognition- Workshops. pp. 49–54. IEEE (2010) 3

    2010

  22. [30]

    IEEE Transactions on Pattern Analysis and Machine Intelligence41, 2222–2235 (2019) 6

    Son, K., Hays, J., Cooper, D.: Solving square jigsaw puzzle by hierarchical loop constraints. IEEE Transactions on Pattern Analysis and Machine Intelligence41, 2222–2235 (2019) 6

    2019

  23. [31]

    In: Proceedings Recognizing Artistic Style of Archaeological Image Fragments 17 of the IEEE conference on computer vision and pattern recognition

    Szegedy, C., Liu, W., Jia, Y., Sermanet, P., Reed, S., Anguelov, D., Erhan, D., Vanhoucke, V., Rabinovich, A.: Going deeper with convolutions. In: Proceedings Recognizing Artistic Style of Archaeological Image Fragments 17 of the IEEE conference on computer vision and pattern ...

    2015

  24. [32]

    In: International conference on machine learning

    Tan, M., Le, Q.: Efficientnet: Rethinking model scaling for convolutional neural networks. In: International conference on machine learning. pp. 6105–6114. PMLR (2019) 2, 9, 11

    2019

  25. [33]

    arXiv preprint arXiv:2410.24010 (2024) 2

    Tsesmelis, T., Palmieri, L., Khoroshiltseva, M., Islam, A., Elkin, G., Shahar, O.I., Scarpellini, G., Fiorini, S., Ohayon, Y., Alali, N., et al.: Re-assembling the past: The repair dataset and benchmark for real world 2d and 3d puzzle solving. arXiv preprint arXiv:2410.24010 (2024) 2

    2024 arXiv

  26. [34]

    Computers & Graphics33(4), 484–495 (2009) 1

    Wallraven, C., Fleming, R., Cunningham, D., Rigau, J., Feixas, M., Sbert, M.: Cat- egorizing art: Comparing humans and computers. Computers & Graphics33(4), 484–495 (2009) 1

    2009

  27. [35]

    Axiomathes15(3), 319–351 (2005) 1

    Willats, J., Durand, F.: Defining pictorial style: Lessons from linguistics and com- puter graphics. Axiomathes15(3), 319–351 (2005) 1

    2005

  28. [36]

    Journal of Visu- alization 24, 251–259 (2021) 3

    Yang, K., Cao, X., Geng, G., Li, K., Zhou, M.: Classification of 3d terracotta warriors fragments based on geospatial and texture information. Journal of Visu- alization 24, 251–259 (2021) 3

    2021

  29. [37]

    Harvard University Press, Cam- bridge, MA (1998), discusses the impact of the eruption of Mount Vesuvius on the preservation of Pompeii 4

    Zanker, P.: Pompeii: Public and Private Life. Harvard University Press, Cam- bridge, MA (1998), discusses the impact of the eruption of Mount Vesuvius on the preservation of Pompeii 4

    1998

Pith tools

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