Pith. sign in

REVIEW 3 major objections 5 minor 60 references

AmalthAI: An Open-Source Computer Vision Platform for Cultural Heritage

T0 review · 3 major / 5 minor · reviewed 2026-08-14 · deepseek-v4-flash

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

Pith's one-line read Textile imprints on clay preserve visual features that trained classifiers can read, and a no-code platform lets archaeologists do the reading.

desk verdict AmalthAI is a genuine open-source platform contribution; the archaeological hypothesis test is honestly framed but not yet leakage-controlled. read the letter →

arxiv 2608.13343 v1 pith:VK6YQ3HV submitted 2026-08-13 cs.CV

Christos Chatzisavvas , Stelios Alvanos , Efstratios Politis , Panagiotis Rigas , Thomas Pappas , Ioannis Giannoukos , Nikolaos Mitianoudis , Agata Ulanowska
show 8 more authors
Katarzyna Żebrowska Nazarij Bu{l}awka Christina Margariti George Pavlidis Chairi Kiourt Anestis Koutsoudis Vassilis Katsouros George Ioannakis
This is my paper · ORCID
classification cs.CV
keywords culturalheritagecomputervisionmachinelearningplatformMLOpsarchaeologytextileimprintsonclaymodelinterpretabilitydatasovereignty
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 tries to establish two linked claims: that textile imprints pressed into clay retain visual features distinguishing raw material (flax, nettle, lime bast, wool) and production technique (drilling, spinning, splicing), and that an archaeologist without machine-learning training can recover that information using a self-hosted, no-code platform. The evidence comes from controlled experimental imprints: a classifier reached 77.53% material accuracy and 85.57% technique accuracy, and segmentation isolated the imprint region at roughly 90% mean intersection-over-union. The authors present these numbers as workflow-validation evidence for the platform rather than as a benchmark claim. If the claims hold, cultural heritage researchers could test archaeological hypotheses about degraded or indirect textile evidence on their own infrastructure, without sending state-owned or restricted data to external cloud services.

What carries the argument

The central mechanism is AmalthAI's closed human-in-the-loop workflow. A web interface hides dataset management, training, and inference behind guided forms; an orchestration layer runs parallel hyperparameter-search trials and keeps the best checkpoint; inference runs in disposable containers built from the same training environment; and Grad-CAM (a heatmap technique showing which image regions drove the prediction) is fed, together with the original image and prediction metadata, to a vision-language model that writes a plain-language account of where the model looked and why it may have erred. The same storage layer can run fully on-site or synchronize with an external data service, which is what makes restricted artifact data usable in the first place.

What would settle it

Take the same experimental materials and images, then retrain and test under a protocol where clay preparation, lighting, and camera settings are systematically crossed with material and technique labels; if accuracy drops to near chance when the trained model is evaluated on imprint classes imaged under unseen clay and lighting conditions, or when those covariates are shuffled, the hypothesis that imprints preserve recoverable textile information would be refuted.

Watch

Extended reading notes

Core claim

On the paper's own terms, the discovery is that experimentally produced clay imprints carry recoverable label-relevant information about the original textile: misclassifications concentrate in classes that archaeologists themselves find visually hard to separate, which the authors read as evidence that the models are exploiting textile-related properties. The supporting design discovery is that a single graphical platform can carry a non-expert from raw images to trained, reusable classifiers and segmenters; repeated train/test splits yielded similar scores, and the best-performing classification model produced the headline accuracies. The stated scope is deliberately modest: the results are consistent with the hypothesis that imprints preserve such information, not proof that every imprint does, and not a replacement for expert analysis.

Load-bearing premise

The claim stands or falls on whether the models' accuracy reflects the textile attributes rather than incidental visual cues — clay texture, lighting, and camera settings — that co-vary with the classes in the experimental images.

Editorial extensions

If this is right

  • Archaeologists without programming skills can train, compare, and reuse models for multiple label schemas from a single interface, so the same platform can test several hypotheses on one corpus.
  • Because all assets can stay on institutional hardware, the workflow is compatible with state-owned or rights-encumbered artifact collections that cannot be uploaded to vendor clouds.
  • The segmentation layer lets subsequent analysis concentrate on the imprint alone, screening out clay texture, lighting, and background irregularities.
  • Misclassification patterns across classes give a first-pass signal of which textile distinctions are visually confusable, and can direct experts to inspect those cases with the explanation loop.
  • Repeated splits with similar scores suggest the results are not an artifact of one favorable data partition, so the platform's default training setup can serve as a repeatable protocol.

Reading between the lines

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

  • Beyond the paper: the strongest test of the archaeological claim would be cross-condition generalization — training on imprints made with one clay preparation and lighting regime, then testing on imprints made with different clay and illumination; a collapse to chance would show the classifiers are reading dataset-specific cues rather than textile properties.
  • Beyond the paper: Grad-CAM plus a vision-language description could be used as a post-hoc shortcut detector for non-experts, but the paper does not measure whether this loop changes expert trust or improves final conclusions.
  • Beyond the paper: the same no-code workflow could be applied to other indirect archaeological evidence, such as tool marks or residues, whenever experts can define classes on archaeological grounds.
  • Beyond the paper: the platform's per-user data isolation and content-addressed storage design point toward a general pattern for privacy-preserving ML in heritage settings, but the paper does not evaluate multi-institution collaborative workflows.
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

3 major / 5 minor

Summary. The paper introduces AmalthAI, an open-source, self-hostable web-based machine learning platform aimed at cultural heritage (CH) researchers without machine learning expertise. The platform supports image classification, semantic segmentation, and object detection, with dataset management, training via Kubeflow/Katib, hyperparameter search, Grad-CAM explanations, and a vision-language assistant. The authors validate the platform on an experimental archaeology case study: textile imprints on clay. They report segmentation mIoU around 90%, material classification accuracy of 77.53%, and production-technique accuracy of 85.57%, with repeated random splits yielding consistent results. They argue these results support the hypothesis that clay imprints preserve visual features informative of the original textile's raw material and production technique, while acknowledging that the results do not by themselves exclude reliance on dataset-specific cues.

Significance. If the platform works as described, it addresses a real accessibility gap: CH researchers often cannot configure, train, and evaluate CV models, and many restricted heritage datasets cannot be uploaded to commercial cloud platforms. The paper's engineering is detailed and credible: the architecture is modular, the code is released, inference is run in ephemeral containers matching the training environment, storage is content-addressed and supports on-premise deployment, and the multi-user/OAuth design is sensible. The repeated-split reporting with means and standard deviations is stronger than a single-split evaluation. However, the central archaeological finding is not yet rigorously established: the case-study dataset spans different clay textures, lighting conditions, and camera settings, and the classification models were trained on full images, leaving open the possibility that accuracy reflects acquisition covariates rather than textile-structure information. The usability evidence also rests on self-evaluation by co-author experts. These issues are fixable, but they currently limit the strength of the paper's headline claims.

major comments (3)
  1. [Sec. 4.3, Table 3, Fig. 5] The headline archaeological conclusion is not secured against covariate leakage. The dataset is described in Fig. 5 as spanning different clay textures, lighting conditions and camera settings, yet the paper reports no per-image metadata, no balance checks across these covariates, and classification is trained on full images rather than on the segmented imprint region. Because a covariate that is stable within a material or technique class survives repeated random splits, the reported 77.53% material and 85.57% technique accuracies could reflect background or acquisition cues rather than textile-structure information in the imprints. The authors' own caveat in Sec. 4.3 ('does not by itself exclude reliance on dataset specific cues') is to the point. I request leakage-controlled evidence: classification restricted to the segmented imprint ROI, cross-clay or cross-lighting generalization experiments, or explicit demonstration that covariates are balanced across classes.
  2. [Sec. 4.2] The usability claim that CH experts with minimal ML knowledge can independently produce and validate findings is supported only by a self-evaluation by co-author experts. The experts who ran the experiments are also authors of the platform and of the prior framework [3]; no independent participant pool, task-completion metrics, time-on-task, error rates, or comparison against a baseline workflow is reported. This is not fatal for a systems paper, but the phrasing 'enabling non-ML CH experts to independently produce and validate archaeologically meaningful findings' in the Abstract overstates the evidence. Please either soften the claim or add a structured usability assessment.
  3. [Sec. 4.1, Sec. 4.3] The hypothesis success condition requires that misclassifications be attributable to structural similarities between textile classes, but the only evidence is qualitative expert opinion. Table 3 shows nettle and splicing have lower accuracies, and the text states these are 'according to CH experts' difficult cases, without a formal confusion-pattern analysis, a human-baseline comparison, or pre-registered criteria. This leaves the link between model errors and domain-valid similarity anecdotal and compounds the confound identified in the previous comment.
minor comments (5)
  1. [Tables 2 and 3] Please clarify why EfficientNetB0 material accuracy is 76.32±1.21 in Table 2 but 77.53% in Table 3; if Table 3 uses the best split or a different test set, state this explicitly and provide variance for the reported inference results.
  2. [Sec. 4.2] The sentence 'with only a few samples outside the defined classes' is vague; report exact counts or explain how such samples were treated in the experiments.
  3. [Sec. 3.4 and Figures 7–10] The VLM system prompt forbids describing materials as 'clay', but the examples in Figures 8 and 10 describe 'clay imprint' and 'clay or similar material'; the prompt-output inconsistency should be resolved or explicitly discussed.
  4. [General] The case-study dataset is not released, so the quantitative results cannot be reproduced independently; given the data-sovereignty motivation this is understandable, but the paper should state explicitly which parts of the case study can be reproduced from the public code and which require the private dataset.
  5. [Sec. 1, contribution 3] The phrase 'demonstrating its practical effectiveness and usability' is asserted from a single case study; consider 'demonstrated in a case study' to avoid overclaiming generalizations about usability.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity found: the archaeological hypothesis and platform claims are evaluated empirically on held-out data, not derived from the model inputs by construction.

full rationale

The paper's central claim is that clay textile imprints retain visual information about raw material and production technique, and that a no-code platform can let cultural heritage experts train models to recover that information. This claim is tested empirically: classification accuracy is measured on held-out test splits after training on curated labels, and segmentation mIoU is measured on expert-annotated masks. The reported numbers (e.g., 77.53% material accuracy, 85.57% technique accuracy, roughly 90% mIoU) are experimental outputs, not quantities that the platform or hypothesis definition forces by construction. No fitted parameter is relabeled as a prediction, and no equation-level identity between input and output is present. The citation to prior work [3] for modularity principles is a normal self-citation and is not load-bearing: the architecture choices described in Section 3 are presented as design decisions, not as a theorem or derivation. The paper's caveat in Section 4.3 that the results 'do not by itself exclude reliance on dataset specific cues' flags a potential confounding problem (e.g., clay texture, lighting, camera settings correlated with class labels), but that is an external-validity or correctness risk, not circularity. The repeated-split results in Table 2 provide some evidence against favorable-partition artifacts, and the authors explicitly frame the results as workflow-validation evidence rather than benchmark claims. Overall, the derivation chain is empirical and self-contained; no load-bearing step reduces to its own inputs, and no self-citation chain forces the conclusion.

Assumptions & free parameters 3 free parameters · 4 assumptions · 0 invented entities

The platform's validation rests on empirical choices rather than mathematical axioms; this ledger records the unverified assumptions in the case study and the hand-picked experimental settings.

free parameters (3)
  • Katib hyperparameter search ranges = LR 0.001-0.1; batch size 4-16; epochs 30-50 (classification); LR 0.001-0.1; batch 8-16; epochs 20-30 (segmentation)
    Chosen ad hoc by the authors for the case study (Table 1); the final model performance depends on these search spaces, but they are not derived from any theory.
  • Data augmentation policies
    Selected through the platform's interface during the case study; the exact transforms are absent, so the augmentation effect on reported accuracy is unstated (Section 4.2).
  • VLM system prompt and in-context examples = Hand-written JSON prompt with four examples (Figs. 6-10)
    The explanation quality of the vision-language assistant is shaped by these hand-crafted prompts; they are free choices of the authors, not benchmarked.
assumptions (4)
  • domain assumption The experimental clay-imprint dataset is representative of archaeological textile imprints and labels are correct.
    The hypothesis extrapolates from controlled experimental imprints to archaeological evidence; the paper treats the curated dataset as ground truth (Sec. 4).
  • domain assumption Held-out classification accuracy on this dataset is not driven by dataset-specific cues such as imaging conditions or near-duplicate samples.
    The paper explicitly acknowledges this possibility in Sec. 4.3; if false, the archaeological hypothesis is unsupported.
  • domain assumption Standard i.i.d. train/test splits on images do not leak across multiple images of the same textile sample.
    The dataset may contain multiple images per controlled sample; repeated random splits do not guarantee sample-level separation. This is not discussed in the paper.
  • domain assumption Grad-CAM and Qwen2-VL outputs are reliable enough for expert review to validate model reasoning.
    The platform's explainability loop assumes these tools provide meaningful evidence (Sec. 3.4); no quantitative validation is provided.

how reviews work

0 comments
Cite this review

Pith. "Pith review of AmalthAI: An Open-Source Computer Vision Platform for Cultural Heritage." pith.science (2026). https://pith.science/paper/VK6YQ3HV

@misc{pith2026260813343,
  author       = {Pith},
  title        = {Pith review of: AmalthAI: An Open-Source Computer Vision Platform for Cultural Heritage},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/VK6YQ3HV}},
  note         = {Machine review of arXiv:2608.13343}
}
read the original abstract

Computer vision (CV) and machine learning (ML) offer new tools for cultural heritage (CH) artifact analysis, but the CV/ML pipeline remains largely inaccessible to CH domain experts, who lack the background to configure, train, or assess models. We present AmalthAI, an open-source CV platform that bridges this gap, enabling non-ML CH experts to independently produce and validate archaeologically meaningful findings. The interface covers dataset management, training, and inference for classification, segmentation, and object detection, with Kubeflow and Katib handling scalable training and hyperparameter search. Grad-CAM localizes the image region behind a prediction, and a vision-language model (VLM) adds a text description of it for expert review. Since archaeological data is often state-owned or rights-encumbered and cannot leave institutional custody, AmalthAI's self-hostable deployment ensures sensitive data is kept within premises. We test the platform on an archaeological use case built on a custom dataset of clay textile imprints, where CH experts trained and validated segmentation, and classification models for hypothesis testing. We provide the implementation code at https://github.com/TEXTaiLES/AmalthAI.

Figures

Figures reproduced from arXiv: 2608.13343 by the authors.

Figure 1
Figure 1. Overview of the AmalthAI architecture and CH computer vision workflow. The platform exposes three supported task modalities (classification, segmentation, and object detection) within a unified interface. A CH expert interacts with the system through three main stages: data management, where image datasets are imported or accessed from a database; model training, where hyperparameter tuning, training, and model sele… view at source ↗
Figure 2
Figure 2. Model training configuration interface. The platform enables intuitive selection of model architectures, hyperparameter tuning, and data augmentations without in￾teraction with underlying code. Inference. The inference module allows users to deploy previously trained mod￾els directly through the web interface. Users may submit either a single image or multiple images simultaneously for batch inference. For every inf… view at source ↗
Figure 3
Figure 3. Stateless inference. For each request, the platform instantiates a temporary Docker container from the same software environment used during training, loads the selected trained model, performs the prediction, returns the results to the interface, and discards the container on completion. 3.4 Vision-Language Assistant In addition to its machine learning capabilities, AmalthAI integrates a Vision￾Language Model (VLM)… view at source ↗
Figures from the paper (9 more)
Figure 4
Figure 4. Figure 4: AmalthAI’s modular data layer. (a) Standalone: assets are read from and writ￾ten to local storage through the storage-abstraction layer; no external service is re￾quired. (b) With an external service, the local directory becomes a cache: assets are rehydrated on first …
Figure 5
Figure 5. Figure 5: Indicatory samples from the experimental textile imprint dataset. Samples have been selected to represent the variety and complexity of the dataset across different clay textures, lighting conditions and camera settings. regarding the examined use case is the following…
Figure 6
Figure 6. Figure 6: System prompt for controlled visual explanations, separating raw-image de￾scription from Grad-CAM attention while preventing class labels, correctness judg￾ments, and unsupported interpretations [PITH_FULL_IMAGE:figures/full_fig_p021_6.png]
Figure 7
Figure 7. Figure 7: Example of a visual explanation pipeline for material-related errors. The model processes a raw microscope image (a), its Grad-CAM heatmap (b), and prediction metadata to produce a structured JSON explanation of a misclassified material [PITH_FULL_IMAGE:figures/full_f…
Figure 8
Figure 8. Figure 8: Example of a visual explanation pipeline for material-related errors. The model processes a raw microscope image (a), its Grad-CAM heatmap (b), and prediction metadata to produce a structured JSON explanation of a misclassified material [PITH_FULL_IMAGE:figures/full_f…
Figure 9
Figure 9. Figure 9: Example of a visual explanation pipeline for errors in technique recognition. A raw microscope image (a), its corresponding Grad-CAM visualization (b), and classifi￾cation outputs are combined to generate a structured JSON explanation of a technique￾level misclassifica…
Figure 10
Figure 10. Figure 10: Example of a visual explanation pipeline for errors in technique recognition. A raw microscope image (a), its corresponding Grad-CAM visualization (b), and classifi￾cation outputs are combined to generate a structured JSON explanation of a technique￾level misclassific…
Figure 11
Figure 11. Figure 11: summarizes the main AmalthAI user-interface visualizations. The dataset interface supports inspection of uploaded data splits and dataset attributes, the inference interface presents model outputs and segmentation masks, and the AI Assistant page provides interaction …
Figure 12
Figure 12. Figure 12: presents qualitative segmentation results for the textile-imprint dataset. These examples complement the quantitative segmentation results by illustrating how the learned masks can support downstream expert inspection of the relevant imprint area. Image Ground Truth I…

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

60 extracted references · 45 canonical work pages

  1. [3]

    In: Computer Applica- tions and Quantitative Methods in Archaeology Conference (CAA2025)

    Chatzisavvas, C., Pappas, T., Rigas, P., Mitianoudis, N., Pavlidis, G., Kiourt, C., Koutsoudis, A., Katsouros, V., Ioannakis, G.: Towards an easy-to-use ma- chine learning framework for cultural heritage scientists. In: Computer Applica- tions and Quantitative Methods in Archaeology Conference (CAA2025). Athens, Greece (May 2026).https://doi.org/10.5281/z...

    work page doi:10.5281/zenodo.20048428 2026

  2. [1]

    In: 12th USENIX symposium on operating systems design and implementation (OSDI 16)

    Abadi, M., Barham, P., Chen, J., Chen, Z., Davis, A., Dean, J., Devin, M., Ghe- mawat, S., Irving, G., Isard, M., et al.: TensorFlow: a system for Large-Scale ma- chine learning. In: 12th USENIX symposium on operating systems design and implementation (OSDI 16). pp. 265–283 (2016) 16 C. Chatzisavvaset al

    work page 2016

  3. [2]

    28, 2026

    Authentik Security Inc.: authentik: The authentication glue you need.https:// goauthentik.io(2026), accessed: Jun. 28, 2026

    work page 2026

  4. [4]

    In: Proceedings of the European conference on computer vision (ECCV)

    Chen, L.C., Zhu, Y., Papandreou, G., Schroff, F., Adam, H.: Encoder-decoder with atrous separable convolution for semantic image segmentation. In: Proceedings of the European conference on computer vision (ECCV). pp. 801–818 (2018)

    2018

  5. [5]

    Computers in Industry134, 103551 (2022)

    Chen, M., Yu, L., Zhi, C., Sun, R., Zhu, S., Gao, Z., Ke, Z., Zhu, M., Zhang, Y.: Improved faster R-CNN for fabric defect detection based on Gabor filter with Genetic Algorithm optimization. Computers in Industry134, 103551 (2022)

    work page 2022

  6. [6]

    28, 2026

    ClearML: Clearml - your entire mlops stack in one open-source tool (2024),https: //clear.ml/, accessed: Jun. 28, 2026

    work page 2024

  7. [7]

    28, 2026

    CVAT.ai Corporation: Computer Vision Annotation Tool (CVAT) (Nov 2023), https://github.com/cvat-ai/cvat, accessed: Jun. 28, 2026

    work page 2023

  8. [8]

    Fibres and Textiles in Eastern Europe15, 64–65 (01 2007)

    Cybulska, M.: Archaeological textiles–a need for new methods of analysis and re- construction. Fibres and Textiles in Eastern Europe15, 64–65 (01 2007)

    work page 2007

Show all 60 references
  1. [9]

    29, 2026

    Directus: Directus: The collaborative backend and headless cms.https : / / directus.com(2026), accessed: Jun. 29, 2026

    2026

  2. [10]

    25, 2026

    Dwyer, B., Nelson, J., Hansen, T., et al.: Roboflow (version 1.0) (2026),https: //roboflow.com, accessed: Jun. 25, 2026

    2026

  3. [11]

    Advances in Archaeological Practice11(2), 152–163 (2023)

    Eberl, M., Bell, C.S., Spencer-Smith, J., Raj, M., Sarubbi, A., Johnson, P.S., Rieth, A.E., Chaudhry, U., Aguila, R.E., McBride, M.: Machine learning–based identifi- cation of lithic microdebitage. Advances in Archaeological Practice11(2), 152–163 (2023)

    2023

  4. [12]

    Journal of Cultural Heritage78, 228–237 (2026)

    ElBehairy, A., El-Nasr, N.A.A., Grimberg, P., Said, L.A.: A comprehensive review of deep learning methods in damage classification, detection, and segmentation of cultural heritage sites. Journal of Cultural Heritage78, 228–237 (2026)

    2026

  5. [13]

    Journal of Computer Applications in Ar- chaeology (Jun 2025).https://doi.org/10.5334/jcaa.190

    Eleftheriadou, A., McPherron, S.P., Marreiros, J.: Machine learning applications in use-wear analysis: A critical review. Journal of Computer Applications in Ar- chaeology (Jun 2025).https://doi.org/10.5334/jcaa.190

    2025 doi

  6. [14]

    Pattern Recognition Letters133, 102–108 (2020)

    Fiorucci, M., Khoroshiltseva, M., Pontil, M., Traviglia, A., Del Bue, A., James, S.: Machine learning for cultural heritage: A survey. Pattern Recognition Letters133, 102–108 (2020)

    2020

  7. [15]

    Hardt,D.:TheOAuth2.0AuthorizationFramework.RFC6749(Oct2012).https: //doi.org/10.17487/RFC6749,https://www.rfc-editor.org/info/rfc6749/

    doi

  8. [16]

    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)

    2016

  9. [17]

    25, 2026

    Intel: Intel Geti (2023),https://github.com/open- edge- platform/geti, ac- cessed: Jun. 25, 2026

    2023

  10. [18]

    Textile research journal92(1-2), 30–42 (2022)

    Jing, J., Wang, Z., Rätsch, M., Zhang, H.: Mobile-Unet: An efficient convolutional neural network for fabric defect detection. Textile research journal92(1-2), 30–42 (2022)

    2022

  11. [19]

    arXiv preprint arXiv:2606.03748 (2026)

    Jocher, G., Qiu, J., Liu, M., Lyu, S., Akyon, F.C., Kalfaoglu, M.E.: Ultra- lytics YOLO26: Unified Real-Time End-to-End Vision Models. arXiv preprint arXiv:2606.03748 (2026)

    2026 arXiv

  12. [20]

    keycloak.org(2026), accessed: Jun

    Keycloak: Keycloak: Open source identity and access management.https://www. keycloak.org(2026), accessed: Jun. 28, 2026 AmalthAI: An Open-Source Computer Vision Platform for Cultural Heritage 17

    2026

  13. [21]

    In: Proceedings of the IEEE/CVF international conference on computer vision

    Kirillov, A., Mintun, E., Ravi, N., Mao, H., Rolland, C., Gustafson, L., Xiao, T., Whitehead, S., Berg, A.C., Lo, W.Y., et al.: Segment anything. In: Proceedings of the IEEE/CVF international conference on computer vision. pp. 4015–4026 (2023)

    2023

  14. [22]

    IEEE access11, 31866–31879 (2023)

    Kreuzberger, D., Kühl, N., Hirschl, S.: Machine learning operations (MLops): Overview, definition, and architecture. IEEE access11, 31866–31879 (2023)

    2023

  15. [23]

    kubeflow.org/, accessed: Jun

    Kubeflow: A machine learning toolkit for Kubernetes (2021),https://www. kubeflow.org/, accessed: Jun. 28, 2026

    2021

  16. [24]

    Applied Physics A106, 309 – 323 (2011)

    Liang, H.: Advances in multispectral and hyperspectral imaging for archaeology and art conservation. Applied Physics A106, 309 – 323 (2011)

    2011

  17. [25]

    In: European conference on computer vision

    Lin, T.Y., Maire, M., Belongie, S., Hays, J., Perona, P., Ramanan, D., Dollár, P., Zitnick, C.L.: Microsoft coco: Common objects in context. In: European conference on computer vision. pp. 740–755. Springer (2014)

    2014

  18. [26]

    IEEE Ac- cess12, 100167–100185 (2024)

    Ling, Z., Delnevo, G., Salomoni, P., Mirri, S.: Findings on machine learning for identification of archaeological ceramics: A systematic literature review. IEEE Ac- cess12, 100167–100185 (2024)

    2024

  19. [27]

    Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) (2022)

    Liu, Z., Mao, H., Wu, C.Y., Feichtenhofer, C., Darrell, T., Xie, S.: A convnet for the 2020s. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) (2022)

    2022

  20. [28]

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

    Long, J., Shelhamer, E., Darrell, T.: Fully convolutional networks for semantic segmentation. In: Proceedings of the IEEE conference on computer vision and pattern recognition. pp. 3431–3440 (2015)

    2015

  21. [29]

    In: Proceedings of the European conference on computer vision (ECCV)

    Ma, N., Zhang, X., Zheng, H.T., Sun, J.: Shufflenet v2: Practical guidelines for efficient CNN architecture design. In: Proceedings of the European conference on computer vision (ECCV). pp. 116–131 (2018)

    2018

  22. [30]

    on the problem of the most efficient tests of statisti- cal hypotheses

    Neyman, J., Pearson, E.S.: Ix. on the problem of the most efficient tests of statisti- cal hypotheses. Philosophical Transactions of the Royal Society of London, Series A: Containing Papers of a Mathematical or Physical Character231(694-706), 289– 337 (02 1933)

    1933

  23. [31]

    Journal of Archaeological Science191, 106607 (2026)

    Orengo, H., Berganzo-Besga, I., Esmoris, J., Lumbreras, F., Aliende, P., Wallace, M., Livarda, A.: High-performance 3D morphometrics via deep learning and tabu- lar foundation models: a case study on complex cereal grain classification. Journal of Archaeological Science191, 10...

    2026

  24. [32]

    Journal of Cultural Heritage74, 214–224 (2025)

    Ottoni, A.L.C., Ottoni, L.T.C.: A deep learning approach for cultural heritage building classification using transfer learning and data augmentation. Journal of Cultural Heritage74, 214–224 (2025)

    2025

  25. [33]

    World Archaeology 40(1), 1–6 (2008)

    Outram, A.K.: Introduction to experimental archaeology. World Archaeology 40(1), 1–6 (2008)

    2008

  26. [34]

    https://palletsprojects.com (2024), accessed: Jun

    Pallets: Flask (version 3.1.0). https://palletsprojects.com (2024), accessed: Jun. 28, 2026

    2024

  27. [35]

    Advances in neural information processing sys- tems32(2019)

    Paszke, A., Gross, S., Massa, F., Lerer, A., Bradbury, J., Chanan, G., Killeen, T., Lin, Z., Gimelshein, N., Antiga, L., et al.: Pytorch: An imperative style, high- performance deep learning library. Advances in neural information processing sys- tems32(2019)

    2019

  28. [36]

    Expert systems with applications239, 122399 (2024)

    Pei, H., Zhang, C., Zhang, X., Liu, X., Ma, Y.: Recognizing materials in cultural relic images using computer vision and attention mechanism. Expert systems with applications239, 122399 (2024)

    2024

  29. [37]

    In: 2020 25th International Conference on Pattern Recognition (ICPR)

    Ragusa, F., Di Mauro, D., Palermo, A., Furnari, A., Farinella, G.M.: Semantic object segmentation in cultural sites using real and synthetic data. In: 2020 25th International Conference on Pattern Recognition (ICPR). pp. 1964–1971. IEEE (2021) 18 C. Chatzisavvaset al

    2021

  30. [38]

    arXiv preprint arXiv:2406.00423 (2024)

    Rei, L., Mladenić, D., Dorozynski, M., Rottensteiner, F., Schleider, T., Troncy, R., Lozano, J.S., Salvatella, M.G.: Multimodal metadata assignment for cultural heritage artifacts. arXiv preprint arXiv:2406.00423 (2024)

    2024 arXiv

  31. [39]

    In: International Conference on Medical image computing and computer-assisted intervention

    Ronneberger, O., Fischer, P., Brox, T.: U-net: Convolutional networks for biomedi- cal image segmentation. In: International Conference on Medical image computing and computer-assisted intervention. pp. 234–241. Springer (2015)

    2015

  32. [40]

    International journal of computer vision115(3), 211–252 (2015)

    Russakovsky, O., Deng, J., Su, H., Krause, J., Satheesh, S., Ma, S., Huang, Z., Karpathy, A., Khosla, A., Bernstein, M., et al.: Imagenet large scale visual recog- nition challenge. International journal of computer vision115(3), 211–252 (2015)

    2015

  33. [41]

    The OpenID Foundation, specification 335(2014)

    Sakimura, N., Bradley, J., Jones, M., De Medeiros, B., Mortimore, C.: Openid connect core 1.0 incorporating errata set 1. The OpenID Foundation, specification 335(2014)

    2014

  34. [42]

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

    Sandler, M., Howard, A., Zhu, M., Zhmoginov, A., Chen, L.C.: Mobilenetv2: In- verted residuals and linear bottlenecks. In: Proceedings of the IEEE conference on computer vision and pattern recognition. pp. 4510–4520 (2018)

    2018

  35. [43]

    In- ternational journal of computer vision128(2), 336–359 (2020)

    Selvaraju, R.R., Cogswell, M., Das, A., Vedantam, R., Parikh, D., Batra, D.: Grad- cam: visual explanations from deep networks via gradient-based localization. In- ternational journal of computer vision128(2), 336–359 (2020)

    2020

  36. [44]

    Journal of Archaeological Method and Theory32(1), 31 (2025)

    Sferrazza, P.: Archaeological and experimental lithic microwear classification through 2D textural analysis and machine learning. Journal of Archaeological Method and Theory32(1), 31 (2025)

    2025

  37. [45]

    International Journal of Computational Intelligence Systems17(1), 11 (2024)

    Sha, S., Li, Y., Wei, W., Liu, Y., Chi, C., Jiang, X., Deng, Z., Luo, L.: Image classi- fication and restoration of ancient textiles based on convolutional neural network. International Journal of Computational Intelligence Systems17(1), 11 (2024)

    2024

  38. [46]

    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)

    2019

  39. [47]

    In: Banck-Burgess, J., Marinova, E., Mischka, D

    Ulanowska,A.:Whynotwool?Evidenceforrawmaterialsandtechnicalusesoftex- tiles based on imprints on the undersides of clay sealings from Bronze Age Greece. In: Banck-Burgess, J., Marinova, E., Mischka, D. (eds.) THE SIGNIFICANCE OF ARCHAEOLOGICAL TEXTILES Papers of the Interna-ti...

    2023

  40. [48]

    25, 2026

    Ultralytics: Ultralytics Platform (2026),https://platform.ultralytics.com/, accessed: Jun. 25, 2026

    2026

  41. [49]

    arXiv preprint arXiv:2409.12191 (2024)

    Wang, P., Bai, S., Tan, S., Wang, S., Fan, Z., Bai, J., Chen, K., Liu, X., Wang, J., Ge, W., et al.: Qwen2-vl: Enhancing vision-language model’s perception of the world at any resolution. arXiv preprint arXiv:2409.12191 (2024)

    2024 arXiv

  42. [50]

    Scientific Re- ports12(1), 13468 (2022)

    Winterbottom, T., Leone, A., Al Moubayed, N.: A deep learning approach to fight illicit trafficking of antiquities using artefact instance classification. Scientific Re- ports12(1), 13468 (2022)

    2022

  43. [51]

    In: International Conference on Computers in Urban Planning and Urban Management

    Yoshimura, Y., Cai, B., Wang, Z., Ratti, C.: Deep learning architect: classification for architectural design through the eye of artificial intelligence. In: International Conference on Computers in Urban Planning and Urban Management. pp. 249–

  44. [52]

    IEEE Data Eng

    Zaharia, M.A., Chen, A., Davidson, A., Ghodsi, A., Hong, S.A., Konwinski, A., Murching, S., Nykodym, T., Ogilvie, P., Parkhe, M., Xie, F., Zumar, C.: Accel- erating the Machine Learning Lifecycle with MLflow. IEEE Data Eng. Bull.41, 39–45 (2018) AmalthAI: An Open-Source Comput...

    2018

  45. [53]

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

    Zhao, H., Shi, J., Qi, X., Wang, X., Jia, J.: Pyramid scene parsing network. In: Proceedings of the IEEE conference on computer vision and pattern recognition. pp. 2881–2890 (2017) 20 C. Chatzisavvaset al. A Supplementary Material A.1 Hardware Requirements & Technical Backgrou...

    2017

  46. [54]

    Image 1 (raw): the actual microscope photograph of the clay imprint

  47. [55]

    Rules:\n

    Image 2 (Grad-CAM): a heatmap overlay showing where the classifier focused. "Rules:\n" "- visible_description: describe ONLY what you see in Image 1. " "Do NOT mention the heatmap or Grad-CAM here.\n" "- attention_description: describe WHERE the heat in Image 2 is spatially " ...

  48. [57]

    visible_description

    The corresponding Grad-CAM visualization. Known classification results (do NOT predict or re-derive them): Ground-truth class: Wool Predicted class: Nettle Prediction confidence: 0.4438 Class probabilities: Flax: 0.37% Nettle: 44.38% Tree bast_lime: 34.35% Wool: 20.90% Using t...

  49. [59]

    visible_description

    The corresponding Grad-CAM visualization. Known classification results (do NOT predict or re-derive them): Ground-truth class: Nettle Predicted class: Wool Prediction confidence: 0.6959 Class probabilities: Flax: 0.25% Nettle: 20.73% Tree bast_lime: 9.43% Wool: 69.59% Using th...

  50. [61]

    visible_description

    The corresponding Grad-CAM visualization. Known classification results (do NOT predict or re-derive them): Ground-truth class: Drilling Predicted class: Spinning Prediction confidence: 0.5679 Class probabilities: Drilling: 37.99% Spinning: 56.79% Splicing: 5.21% Using the orig...

  51. [62]

    The original Cultural Heritage image

  52. [63]

    visible_description

    The corresponding Grad-CAM visualization. Known classification results (do NOT predict or re-derive them): Ground-truth class: Drilling Predicted class: Splicing Prediction confidence: 0.3488 Class probabilities: Drilling: 31.39% Spinning: 33.73% Splicing: 34.88% Using the ori...

Pith tools

Reviewed August 14, 2026 · model on record in the stance chip above.