3D Dataset of Cuneiform Bearing Objects at the Sulaymaniyah Museum

3D Dataset of Cuneiform Bearing Objects at the Sulaymaniyah Museum | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF data-descriptor 3D Dataset of Cuneiform Bearing Objects at the Sulaymaniyah Museum Luis Sáenz, Steffen Bauer, Hubert Mara, Mark Altaweel, Shai Gordin This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-8661845/v1 This work is licensed under a CC BY 4.0 License Status: Posted Version 1 posted You are reading this latest preprint version Abstract This paper presents high-resolution 3D models of cuneiform tablets from the Sulaymaniyah Museum, digitized using the Hexagon SmartScan 3D with approximately 10µm spatial resolution. The dataset includes 44 publishable tablets spanning from 2330 − 330 BCE, processed with GigaMesh open-source software. The 3D models enable advanced research applications including OCR development, fingerprint analysis, improved philological editions, and seal extraction. Enhanced visualization using Multi-Scale Integral Invariants (MSII) filtering reveals previously unreadable text and fine details. This initiative contributes significantly to digital palaeography and cultural heritage preservation while following established criteria for 3D data publication in Assyriology. Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Figure 6 Figure 7 Introduction The cuneiform script, likely invented by the Sumerians in the second half of the fourth millennium BCE in southern Mesopotamia (modern-day southern Iraq), represents one of the earliest systems of writing [1]. This writing system was in use until the 1st century CE, with the most recent known cuneiform tablet dated to 75 CE [2] . Cuneiform script spread throughout the Ancient Near East, leading to its presence in modern-day Turkey, Egypt, Iraq, Iran, Syria, Lebanon, Israel, and other regions. The content of cuneiform tablets is diverse, encompassing a wide range of text genres. These include administrative and judicial documents, as well as scholarly texts such as medical, divinatory, and liturgical works. Additionally, cuneiform tablets feature literary compositions and monumental inscriptions [3]. To create cuneiform signs, scribes made impressions on wet clay tablets using a stylus [4]. Compared to papyrus or leather these tablets are notably thicker. The size and format of the tablets varied according to the time-period, genre, and location. Due to the plasticity of wet clay, not only the script but also seals (see Figure 6 and Figure 7 ) and unintentional fingerprints (see Figure 1 ) were impressed on the tablets. Consequently, the genre, form, depth of the wedges, lighting conditions, and shadows all influence the reading of cuneiform texts. Therefore, the most appropriate digitization method for cuneiform tablets is through 3D scanning. Cuneiform tablets have been found at numerous sites throughout the Near East [3]. Today, many excavated tablets are stored in various museums worldwide. Some museums have digitized their collections or are in the process of doing so through the Cuneiform Digital Library Initiative (CDLI) 1 and the electronic Babylonian Library (eBL). 2 A common preservation method involves creating hand copies of the tablets. Digital preservation methods include digitizing these hand copies and using 2D photography of the tablets. These digital methods are among the most common and cost-efficient approaches to preservation. The cuneiform script was adapted to multiple languages, including from the Semitic and Indo-European language families, but also other isolated languages. Of particular interest in Ancient Mesopotamia are Sumerian and Akkadian. Sumerian, an isolated agglutinative language, is attested in texts from the 28th century BCE until the end of the cuneiform era around the transition to the common era [5]. However, around the second half of the 2nd millennium BCE, Sumerian gradually ceased to be a spoken language and transitioned into a scholarly and liturgical language [6]. The second language of particular interest is Akkadian, a Semitic language originating in northern Mesopotamia. Akkadian is first attested around the 26th century BCE and remained in use until the end of the cuneiform culture, with the aforementioned tablet from 75 CE being written in Akkadian [7]. Although still debated, it is believed that by the mid-5th century BCE, Akkadian began to be replaced by Aramaic as a spoken language, while Akkadian persisted as a written and scholarly language [8]. During this period, the cuneiform script was also gradually replaced by the Aramaic script. The dataset published here contains 3D models of cuneiform tablets from the Slemani Museum, located in the Kurdistan Region of Iraq. This collection hosts more than 6,000 tablets and other cuneiform bearing artefacts (like bricks, prisms, cones, etc.) from various time periods and provenances, with only about 10% of them published. This initiative aimed to scan as many cuneiform tablets and inscribed objects as possible within a ten-day period. To enable a comprehensive analysis of the forms, layouts, and scripts from different historical periods, we selected tablets and objects spanning various chronological phases for scanning. In total, 92 artifacts featuring cuneiform texts in Sumerian and Akkadian were scanned. The distribution of the types of objects is as follows: Cuneiform tablets 87 Brick 1 Clay Hand 1 Sumerian Stela 1 Prisms 2 Total 92 Classifying them by genre results in the following distribution: Archival Texts 87 Literary Texts 1 Monumental 4 Total 92 Archival texts, the most frequently attested genre in the dataset, include letters, payment receipts, delivery documents, and judicial records. However, some literary and monumental texts are also present. Due to publication rights we can only publish in this dataset 44 cuneiform tablets. The distribution of these texts according to genre and subtype is as follow: Archival Texts 41 letters 18 administrative 20 legal 3 Literary Texts 1 Monumental 2 Total 44 Computational Analysis This dataset is being integrated into the CuBeDa (Cuneiform Benchmark Dataset) series, which is essential for training neural networks in the development of e.g. 3D cuneiform OCR (Optical Character Recognition). Initial steps in this direction are discussed in [13]. For OCR development, the 3D models can be annotated as shown in [14]. Additionally, 2D images generated from these 3D models can be annotated, as demonstrated in [10]. These images can also be employed for cuneiform sign recognition, as recently shown by Mikulinsky et al. [15] who demonstrate how prototype alignment techniques can enhance recognition accuracy by leveraging structural configurations, particularly beneficial for rare signs. Moreover, this dataset supports the comparative analysis of 3D versus 2D geometric morphometrics of cuneiform signs across different periods, contributing to the identification of traits in digital palaeography. Fingerprints Analysis This dataset also has significant potential for fingerprint analysis. Some of the tablets exhibit fingerprints, and their extraction is enhanced by MSII (see Figure 1). Such analysis could provide insights into the age and sex of the scribes, offering a deeper understanding of the people who created these ancient documents. A recent study conducted by Jon Ross et al [16] presented new biometric evidence from fingerprints preserved on ceramic objects from the Late Bronze Age II stratum at Tel Burna, providing insights into the demographics of pottery production. The analysis revealed that female potters predominantly produced ceramics. Philological Editions Until now, 3D models have not been particularly useful for philological editions. This can be attributed to various factors, such as a lack of interest in the method or the restricted availability to only a select group of scholars. However, as digital methods become more widespread, the use and inclusion of 3D datasets are adding new levels of quality and possibilities. Processing with GigaMesh and the application of MSII further enhance their potential for reuse in classical philological editions. Consequently, these high-resolution 3D models are emerging as a valuable tool for philological analysis, complementing traditional photographs and handmade copies. As example, Karen Radner [17] published a photograph of a badly damaged prism fragment, SM 043619 (T-7619), and tentatively identified it as part of Esarhaddon Episode 24: Nin. A, VI 60 [18] or Esarhaddon 1 VI 60 in the Oracc (Open Richly Annotated Cuneiform Corpus) edition, based on the clarity of only the penultimate line (see Figure 2). However, after processing the 3D model and applying MSII filtering, it became possible to more clearly identify additional lines, leading to a definitive identification as Esarhaddon 1 III 27-45 (see Figures 3, 4, 5) according to the Oracc classification. Although this fragment does not introduce significant variants to the already reconstructed inscription, aside from line breaks, it demonstrates the potential of this method in enhancing textual analysis. Seals extraction Using different coloration after applying the MSII stamp and cylinder seals become more visible (see Figures 6 and 7). Method The dataset follows a set of criteria for the publication of 3D models in Assyriology, established by Homburg et al. [9] and further applied by Homburg et al. [10] in their work on annotated 3D models of cuneiform tablets, and also Homburg et al. [11] in their Pipeline for Processing the tablet from Haft Tappeh. The 3D acquisition was conducted by Luis Sáenz and Hemin Fatah at the museum, with technical support from Steffen Bauer (Germany). Scanning took place over a ten-day period at the end of September 2023, using a Hexagon SmartScan structured light scanner (SLS) within the museum’s facilities. For cuneiform tablets we used the so-called small field of view with a diagonal of 125mm (S-125). With these settings, the scanner achieves a resolution of around 40µm in the plane (X-Y) and a resolution limit of 5µm in height (Z). According to the manufacturer's specification sheet, the feature accuracy is 9µm. No color information was captured. This configuration is particularly well suited for small tablets, approximately up to 10 cm² in size, which could be fully covered within the 125 mm field of view and typically required 15–20 minutes to scan. Each small tablet typically required 6–8 scan layers. In contrast, larger tablets demanded significantly more time due to the need for multiple overlapping scans. Using lenses with a wider field of view was not a viable solution, as it would have resulted in reduced resolution and, consequently, less accurate 3D models. To improve efficiency, we introduced a measuring procedure using a ruler to define scanning sections of approximately 11 cm², ensuring an overlap of about 1 cm between adjacent scans. This method allowed us to reduce the number of redundant scans and streamline the scanning process for larger tablets. All layers were spatially aligned and subsequently optimized before being merged into a single 3D mesh. Data Record The 3D acquisition of the tablets was conducted using the aforementioned scanner at the museum’s facilities in late September 2023. Immediately after each scan, the resulting 3D model was temporarily stored in the scanner’s proprietary format. At the end of each day, the models were exported as PLY files and processed in GigaMesh, following these steps: Automatic Mesh Polishing Cleaning (Erosion) Filling (Dilation) Manual orientation to establish the artifact’s coordinate system Application of the Multi-Scale Integral Invariants (MSII) to visualize and extract the smallest details Export to the PLY format For steps 1 and 2, a video tutorial available on the Heidelberger Dokumentserver and on YouTube via the GigaMesh channel . A tutorial for step 3 is also available on the same YouTube channel. The primary data product in this publication consists of exported Stanford Polygon (PLY) files. These files are named with the postfix 'GMOCF,' indicating that they have been processed with G iga M esh, o riented, c leaned, and had holes f illed. Orientation defines the front of the tablet, which in the context of a 3D model may differ from the traditional philological designations of 'obverse' and 'reverse.' For example, the front in a 3D model might be what is traditionally considered the reverse if the obverse is significantly damaged. The cleaning process involves removing outliers, zero-area vertices, etc, while filling holes, which are typically small missing sections in the 3D surface. These holes, although often not easily visible, can impact the results of the MSII filter. Additionally, the cleaning and filling process ensures that the models are watertight, making them suitable for 3D printing. GigaMesh also enriches the mesh with metadata, particularly indicating which vertices were artificially added during the filling process. Additionally, metadata about the meshes are exported into CSV (see “Quality Control”). Subsequently, six images per 3D model were rendered, depicting the front, back, top, bottom, left, and right sides of the cuneiform tablets. Each view was individually saved as a PNG file. These images were then compiled into a single-page PDF in the fatcross view using LaTeX. Following this, we applied the MSII filter to visualize fine details, such as fingerprints, damaged cuneiform signs and seal impressions (see Figures 1,5,6 and 7). These details are crucial for further studies (see “Reuse Potential”). The processed 3D models are available for visualization and download via heidICON, a digital image database maintained by Heidelberg University that hosts a wide range of visual research materials, including photographs, drawings, manuscripts, and other multimedia. Within heidICON, the models are published under the Propylaeum Pool, specifically in the subrepository “Keilschrifttafeln aus Sulaymaniyah”. The 3D models are enriched with metadata in accordance with the standards of the Heidelberg University Library. The metadata concerning the quality assessments of the models (see section “Quality Control”) is published in heidDATA, the university’s institutional repository for open research data, and is linked to the corresponding entry in the Propylaeum Pool. These repositories follow the principles of the Heidelberg Cuneiform Benchmark Dataset (HeiCuBeDa), a subrepository dedicated to 3D models of cuneiform and related artifacts from the Hilprecht Collection in Jena. HeiCuBeDa, initiated by Hubert Mara, is also hosted on heidDATA [12]. This work can therefore be summarized in the following steps: (i) acquisition of archaeological artifacts, (ii) generation of 3D models, (iii) mesh cleaning, (iv) manual orientation, (v) rendering of 2D images for each tablet’s side, employing a normalized grayscale and consistent resolution across the entire dataset, (vi) standardized visualization of these images in a PDF using the fatcross rendering method, (vii) application of the 3D MSII filter, and (viii) enrichment of the metadata. Spatial coverage The tablets are stored and were scanned at: Sulaymaniyah Museum, Sulaymaniyah, Kurdistan Region of Iraq. Northern boundary: 35.5579703 Southern boundary: 35.5570720 Eastern boundary: 45.42572153 Western boundary: 45.4257376 The provenance of most of the tablets is largely unknown. However, some tablets are confirmed to originate from cities such as Sippar, Larsa, Adab, Isin and Ninive. These locations suggest that the tablets likely originate from both northern and southern Mesopotamia: Northern boundary: 36.3667° N, 43.1500° E (Near the ruins of Nineveh, close to modern-day Mosul, Iraq) Southern boundary: 31.2267° N, 45.8667° E (Near the ancient city of Larsa, close to modern-day Tell as-Senkereh, Iraq) Western boundary: 33.0608° N, 44.2544° E (Near the ancient city of Sippar, close to modern-day Abu Habbah, Iraq) Eastern Boundary: 32.1133° N, 45.5000° E (Near the ancient city of Adab, close to modern-day Bismaya, Iraq) Temporal coverage The total temporal span of the tablets ranges from 2330 to 330 BCE. The distribution of tablets across different historical periods is as follows: Sargonic (2334–2154 BCE): 7 tablets Ur III (2110–2003 BCE): 16 tablets Neo-Assyrian (1000–600 BCE): 2 tablets Neo-Babylonian (1025–330 BCE): 1 tablet Sampling strategy To comprehensively analyze the forms, layouts, and scripts across different periods, we selected tablets from various chronological phases for scanning. This approach allows us to capture a broad spectrum of variations in tablet format and writing styles. By including samples from multiple periods, we aim to identify and compare the distinct characteristics and developments in the tablets' physical and textual features throughout history. Data type Processed data (3D meshes, PNG and PDF) and interpretation of data (transliteration, translation and quality assessments in CSV-Format). Format names and versions Polygon File Format (PLY), Portable Document Format (PDF), Portable Network Graphic (PNG), Comma-Separated Values (CSV) Creation dates 23/09/2023 – 02/10/2023: 3D scans of cuneiform tablets and other archaeological artifacts at the Sulaymaniyah-Museum and generation of the 3D models. 11/2023: Creation of PLY, PDF and PNG files. 26/04/2025: Creation of CSV files. Dataset Creators Luis Sáenz, performed the 3D scans and exported them in PLY Format, processed the models as described in Method the orientation, created the Metadata and the png renderings of the 3D models as well as the PDF rendetrings in fatcross view. Hemin Nure Fatah, Support and help for creation of the 3D scans at the Museum Steffen Bauer, Supported scanner and Scanner Software, Created the 3D renderings and quality assessment metadata of all cuneiform 3D scans. Nyan Naser Hama Hasan, prepared the tablets for scanning Hashim Hama Abdulla, Coordinated the scanning process at the Museum. Language English in the metadata. Akkadian and Sumerian on the scanned tablets. License CC-BY SA 4.0 Repository location The data is published on heidICON, the repository of the University Library of heidelberg, under the following URL https://heidicon.ub.uni-heidelberg.de/pool/cuneiform_sulaimaniya. The cuneiform tablet scans are published there using the following DOI: SM.036042: https://doi.org/10.11588/heidicon/23958493 SM.036311: https://doi.org/10.11588/heidicon/23958494 SM.036359: https://doi.org/10.11588/heidicon/23958495 SM.036373: https://doi.org/10.11588/heidicon/23958496 SM.036389: https://doi.org/10.11588/heidicon/23958497 SM.036390: https://doi.org/10.11588/heidicon/23958498 SM.036413: https://doi.org/10.11588/heidicon/23958499 SM.036425: https://doi.org/10.11588/heidicon/23958500 SM.036432: https://doi.org/10.11588/heidicon/23958501 SM.036475: https://doi.org/10.11588/heidicon/23958502 SM.036486: https://doi.org/10.11588/heidicon/23958503 SM.036497: https://doi.org/10.11588/heidicon/23958504 SM.036547: https://doi.org/10.11588/heidicon/23958505 SM.037023: https://doi.org/10.11588/heidicon/23958506 SM.037244: https://doi.org/10.11588/heidicon/23958507 SM.037315: https://doi.org/10.11588/heidicon/23958508 SM.037316: https://doi.org/10.11588/heidicon/23958509 SM.037317: https://doi.org/10.11588/heidicon/23958510 SM.037318: https://doi.org/10.11588/heidicon/23958511 SM.037319: https://doi.org/10.11588/heidicon/23958512 SM.037321: https://doi.org/10.11588/heidicon/23958513 SM.037322: https://doi.org/10.11588/heidicon/23958514 SM.037323: https://doi.org/10.11588/heidicon/23958515 SM.037324: https://doi.org/10.11588/heidicon/23958516 SM.037326: https://doi.org/10.11588/heidicon/23958517 SM.037327: https://doi.org/10.11588/heidicon/23958518 SM.037377: https://doi.org/10.11588/heidicon/23958519 SM.037447: https://doi.org/10.11588/heidicon/23958520 SM.037798: https://doi.org/10.11588/heidicon/23958521 SM.037942: https://doi.org/10.11588/heidicon/23958522 SM.039043: https://doi.org/10.11588/heidicon/23958523 SM.039055: https://doi.org/10.11588/heidicon/23958524 SM.039085: https://doi.org/10.11588/heidicon/23958525 SM.039156: https://doi.org/10.11588/heidicon/23958526 SM.039816: https://doi.org/10.11588/heidicon/23958527 SM.039833: https://doi.org/10.11588/heidicon/23958528 SM.039886: https://doi.org/10.11588/heidicon/23958529 SM.039941: https://doi.org/10.11588/heidicon/23958530 SM.039977: https://doi.org/10.11588/heidicon/23958531 SM.041015: https://doi.org/10.11588/heidicon/23958532 SM.041016: https://doi.org/10.11588/heidicon/23958533 SM.041017: https://doi.org/10.11588/heidicon/23958534 SM.043618: https://doi.org/10.11588/heidicon/23958535 SM.043619: https://doi.org/10.11588/heidicon/23958536 The CSV tables are there using the following DOI: https://doi.org/10.11588/DATA/4FROK0. Publication date The dataset in heidICON was published on heidICON on 27/02/2025, the dataset in heiDATA on 09/05/2025. Constraints One of the primary constraints encountered in this initiative is the reluctance of museums to grant publication rights for the 3D models. As a result, we are limited to publishing 3D models of tablets that have already been made publicly available in previous publications. This restriction significantly impacts the breadth of material we can share. Technical Validation The 3D models generated in this study are of high quality, offering both accuracy and reliability for subsequent analysis. To ensure and document these standards, GigaMesh was employed not only for mesh processing but also for exporting comprehensive technical metadata (see “Repository location”). Despite the small physical dimensions of the tablets, the high-resolution scans result in a substantial number of vertices—ranging from 1.1 million to 10.6 million per model, with an average of approximately 2.7 million. A key metric indicating the completeness of the 3D scans is the proportion of synthetic vertices introduced during post-processing in GigaMesh. This proportion remains low, averaging just 1.8%, with a maximum of about 4%. Notably, nearly two-thirds of all models contain fewer than 2% synthetic vertices. Occasional artifacts, such as wave-like distortions in localized scan segments, may occur due to the structured light scanning method, particularly with the specific scanner employed (cf. models SM 039043, SM 037244). These should not be mistaken for intentional imprints such as fingerprints or handprints (see Figure 1). Another rare but observable artifact involves the exaggeration of geometric edges caused by strong color contrasts, especially where inventory numbers are written in black ink on the clay tablets (e.g., SM 037318, SM 036425). Importantly, these infrequent artifacts do not compromise the overall quality or usability of the 3D models for further research and analytical applications. Declarations Data Availability The data is published on heidICON, the repository of the University Library of heidelberg, under the following URL https://heidicon.ub.uni-heidelberg.de/pool/cuneiform_sulaimaniya. The CSV tables are there using the following DOI: https://doi.org/10.11588/DATA/4FROK0. Acknowledgements We would like to express our gratitude to the Sulaymaniyah Museum for their invaluable support and collaboration. We also extend our sincere thanks to Timo Homburg for developing a parsing script that converted the metadata from CSV tables into JSON files, the required format for repository submission. Author Contributions L.S., M.A., and Sh.G. conceived the study, M.A. and Sh.G. supervised and funded the project. L.S. and S.B. were in charge of data creation and curation. S.B. and H.M. were responsible for the methodology, and H.M. created the GigaMesh software used for 3D model analysis. L.S. prepared the first draft of the manuscript with all figures, as well as the initial analysis of 3D models. All authors participated in writing, editing, and revising the manuscript. Funding Statement This initiative has been funded by the Paleography Powered by Machine Learning: Analyzing Cuneiform Documents. LMU-TAU Cooperation Fund, internal grant, and by the Babylonian Industrialism in the Age of Empires. Internal Ariel University grant funding RA2300000111. References Krebernik, M. Zur Struktur und Geschichte des älteren sumerischen Onomastikons, in Alter Orient und Altes Testament 296 1-74 (Münster Ugarit-Verlag 2002). Sachs, A. J. The latest datable cuneiform tablets, in Alter Orient und Altes Testament 25 , 379-398 (Kevelaer Neukirchener Verlag 1976) Streck, M. P. Großes Fach Altorientalistik: Der Umfang des keilschriftlichen Textkorpus, in Mitteilungen der Deutschen Orient-Gesellschaft 142 , 35-58, (2010). Glassner, J.-J. The invention of cuneiform: writing in Sumer (Baltimore Johns Hopkins University Press 2003). Sallaberger, W. Das Ende des Sumerischen: Tod und Nachleben einer altmesopotamischen Sprache, in Münchner Forschungen zur historischen Sprachwissenschaft 2 , 108–140 (Bremen Hempen Verlag 2004). Maul, S. M. Küchensumerisch oder hohe Kunst der Exegese? Überlegungen zur Bewertung akkadischer Interlinearübersetzungen von Emesal-Texten, in Alter Orient und Altes Testament 247 , 253-267 (Neukirchen-Vluyn Neukirchener Verlag 1997). Streck, M. P. Akkadian and Cuneiform in Handbuch der Orientalistik 152 , 66-74 (Leiden, Boston Brill 2021). Hackl, J. The Death of Akkadian as a Written and Spoken Language, in Handbuch der Orientalistik 152 , 1459-1477 (Leiden, Boston Brill 2021). Homburg, T., Cramer, A. , Raddatz, L. & Mara, H. Metadata schema and ontology for capturing and processing of 3D cultural heritage objects https://www.nature.com/articles/s40494-021-00561-w (2021). Homburg, T. Zwick, R., Mara, H. & Bruhn, K.-C. Annotated 3D-Models of Cuneiform Tablets https://openarchaeologydata.metajnl.com/articles/10.5334/joad.92 (2022). Homburg, T., Zwick, R., Bruhn, K.-C. & H. Mara 3D Data Derivatives of the Haft Tappeh Processing Pipeline https://cdli.mpiwg-berlin.mpg.de/articles/cdlj/2022-1 (2022). Mara, H. & Bogacz, B. Breaking the Code on Broken Tablets: The Learning Challenge for Annotated Cuneiform Script in Normalized 2D and 3D Datasets https://ieeexplore.ieee.org/document/8978050 (2019). Stötzner, E., Homburg, T. & Mara, H. CNN based Cuneiform Sign Detection Learned from Annotated 3D Renderings and Mapped Photographs with Illumination Augmentation https://arxiv.org/abs/2308.11277 (2023). Stötzner, E., Homburg, T. , Bullenkamp, J. P. & Mara, H. R-CNN based PolygonalWedge Detection Learned from Annotated 3D Renderings and Mapped Photographs of Open Data Cuneiform Tablets https://diglib.eg.org/items/e9e8eb87-cb45-4c57-b2bc-44fb347b8c4d (2023). Mikulinsky, R., Alper, M., Gordin, S., Jiménez, E., Cohen, Y., and Averbuch-Elor, H. ProtoSnap: Prototype alignment for cuneiform signs in https://openreview.net/forum?id=XHTirKsQV6 (2024). Ross, J., Fowler, K. D., and Shai, I. New Fingerprint Evidence for Female Potters in Late Bronze Age Canaan: The Demographics of Potters and Division of Labour at Tel Burna doi: https://www.sciencedirect.com/science/article/abs/pii/S0278416523000491?via%3Dihub (2023). Radner, K. Cuneiform Inscriptions in the Archaeological Museum of Sulaimaniya Altorientalische Forschungen 52 , 98-103 (2011). Borger, R. Die Inschriften Asarhaddons Königs von Assyrien (Graz Biblio-Verlag 1956). Footnotes https://cdli.earth. https://www.ebl.lmu.de. Additional Declarations No competing interests reported. Cite Share Download PDF Status: Posted Version 1 posted You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-8661845","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"data-descriptor","associatedPublications":[],"authors":[{"id":607240565,"identity":"3a5ab532-e9ec-4510-9704-4055e8266965","order_by":0,"name":"Luis Sáenz","email":"","orcid":"","institution":"Saxon Academy of Sciences in Leipzig","correspondingAuthor":false,"prefix":"","firstName":"Luis","middleName":"","lastName":"Sáenz","suffix":""},{"id":607240566,"identity":"28a71f26-1099-4f56-a1c5-13615e7568d5","order_by":1,"name":"Steffen Bauer","email":"","orcid":"","institution":"Heidelberg University","correspondingAuthor":false,"prefix":"","firstName":"Steffen","middleName":"","lastName":"Bauer","suffix":""},{"id":607240567,"identity":"826bb209-2441-4ab2-9bed-c0910db4ffba","order_by":2,"name":"Hubert Mara","email":"","orcid":"","institution":"Freie Universität Berlin","correspondingAuthor":false,"prefix":"","firstName":"Hubert","middleName":"","lastName":"Mara","suffix":""},{"id":607240568,"identity":"68a596ec-fc03-4fd9-a1e4-199e2601eaa7","order_by":3,"name":"Mark Altaweel","email":"","orcid":"","institution":"University College London","correspondingAuthor":false,"prefix":"","firstName":"Mark","middleName":"","lastName":"Altaweel","suffix":""},{"id":607240569,"identity":"3b3ecd67-b966-49c9-b7c0-cb026b3a3040","order_by":4,"name":"Shai Gordin","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA20lEQVRIie3RMQrCMBiG4a8E6lJ0VRB7hTi5aL1KS0EXB8cODoJQN+eA3sIL/BJwKroKOtgbtCDiJLaKuJmMgnmHEEIe8kMAk+kXqwMEdAEGi8oDu6JHBiXBizANUiSf65NARdzlLN1cor3bmTMiTDxUVYSftlw2k2N7JW2fsA3Vg/G6z2UjPlqCOZxgk5q4YpgVZNcXrJYR7hoEhxHf5DEFxSsgK9Yg/DAaSyRhKJjNKViEjs5g6/wWeT1Rk+k5u3otdz5VTVZ8hPPe+YDz5eIn66Z1zWQymf62B/LnP8ZsVpObAAAAAElFTkSuQmCC","orcid":"","institution":"Ariel University","correspondingAuthor":true,"prefix":"","firstName":"Shai","middleName":"","lastName":"Gordin","suffix":""}],"badges":[],"createdAt":"2026-01-21 16:08:16","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-8661845/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-8661845/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":104919149,"identity":"c94a1343-74b5-40cc-8e5b-1e7e741d7942","added_by":"auto","created_at":"2026-03-18 17:00:05","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":461350,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003e2D rendering of the 3D model of SM 039043, generated using GigaMesh with MSII filtering and an inverted grayscale colormap. The feature vector element is set to 13. A fingerprint is visible on the bottom left edge.\u003c/strong\u003e\u003c/p\u003e","description":"","filename":"floatimage1.png","url":"https://assets-eu.researchsquare.com/files/rs-8661845/v1/a488ece560caa962a13ca4d2.png"},{"id":105034661,"identity":"382c77d6-7da6-4789-8603-829b45ae47de","added_by":"auto","created_at":"2026-03-20 07:23:47","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":779354,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003ePhotograph of SM 043619, as published by Karen Radner [17]. Due to the poor state of preservation, only the penultimate line was legible, leading to a tentative identification as Esarhaddon Episode 24: Nin. A, VI 60.\u003c/strong\u003e\u003c/p\u003e","description":"","filename":"floatimage2.png","url":"https://assets-eu.researchsquare.com/files/rs-8661845/v1/a54ce61e7279d2fff4db57ac.png"},{"id":104919146,"identity":"7f7d0146-769f-406f-9738-25a7cb893711","added_by":"auto","created_at":"2026-03-18 17:00:05","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":309095,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003e2D rendering of the 3D model of SM 043619, generated using GigaMesh with a solid color colormap and M light. Even before applying any additional filtering, this rendering reveals additional lines that can be read more clearly.\u003c/strong\u003e\u003c/p\u003e","description":"","filename":"floatimage3.png","url":"https://assets-eu.researchsquare.com/files/rs-8661845/v1/28d778e8122ec1c9b69a37c0.png"},{"id":105034766,"identity":"426cd9d2-950b-4245-a4c1-2064120af0ec","added_by":"auto","created_at":"2026-03-20 07:24:07","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":313035,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003e2D rendering of the 3D model of SM 043619, generated using GigaMesh with MSII filtering in a hot colormap. The incisions of the cuneiform script are more clearly recognizable in this rendering.\u003c/strong\u003e\u003c/p\u003e","description":"","filename":"floatimage4.png","url":"https://assets-eu.researchsquare.com/files/rs-8661845/v1/83e4abb3b2b53c1aa6948c8e.png"},{"id":105034446,"identity":"5552ad6d-2a1d-4500-8437-0f52505599f2","added_by":"auto","created_at":"2026-03-20 07:23:19","extension":"png","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":381524,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003e2D rendering of the 3D model of SM 043619 made by GigaMesh with MSII filtering in inverted Grayscale colormap. Additionally, the feature vector element is set to 13.\u003c/strong\u003e\u003c/p\u003e","description":"","filename":"floatimage5.png","url":"https://assets-eu.researchsquare.com/files/rs-8661845/v1/dd48bade209fa1bf6c18d4f8.png"},{"id":105034648,"identity":"9bb1c626-c70e-4988-9f7f-27dc7b096f73","added_by":"auto","created_at":"2026-03-20 07:23:45","extension":"png","order_by":6,"title":"Figure 6","display":"","copyAsset":false,"role":"figure","size":858682,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003e2D rendering of the 3D model of the obverse side of SM 039043, generated using GigaMesh with MSII filtering and an inverted Spectral (Brewer, diverging) colormap. Two seal inscriptions are visible, which were only partially copied in the philological edition.\u003c/strong\u003e\u003c/p\u003e","description":"","filename":"floatimage6.png","url":"https://assets-eu.researchsquare.com/files/rs-8661845/v1/d4d514d62c481741e426732a.png"},{"id":104919152,"identity":"36e752ed-4ada-4abe-b694-bbe67cf5818d","added_by":"auto","created_at":"2026-03-18 17:00:05","extension":"png","order_by":7,"title":"Figure 7","display":"","copyAsset":false,"role":"figure","size":800082,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003e2D rendering of the 3D model of the reverse side of SM 039043 90° rotated, generated using GigaMesh with the same configuration as fig. 6. Two same two seals inscriptions of the obverse are visible, which were not included in the philological edition.\u003c/strong\u003e\u003c/p\u003e","description":"","filename":"floatimage7.png","url":"https://assets-eu.researchsquare.com/files/rs-8661845/v1/ec807fcbadf050928715e049.png"},{"id":105037618,"identity":"6c72e915-435f-4fad-8c4a-6d0eb55c239d","added_by":"auto","created_at":"2026-03-20 07:39:54","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":5418451,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-8661845/v1/9a0b397c-382b-42f3-99df-426794b825ca.pdf"}],"financialInterests":"No competing interests reported.","formattedTitle":"3D Dataset of Cuneiform Bearing Objects at the Sulaymaniyah Museum","fulltext":[{"header":"Introduction","content":"\u003cp\u003eThe cuneiform script, likely invented by the Sumerians in the second half of the fourth millennium BCE in southern Mesopotamia (modern-day southern Iraq), represents one of the earliest systems of writing [1].\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003eThis writing system was in use until the 1st century CE, with the most recent known cuneiform tablet dated to 75 CE\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003e[2]\u003cstrong\u003e.\u0026nbsp;\u003c/strong\u003eCuneiform script spread throughout the Ancient Near East, leading to its presence in modern-day Turkey, Egypt, Iraq, Iran, Syria, Lebanon, Israel, and other regions.\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe content of cuneiform tablets is diverse, encompassing a wide range of text genres. These include administrative and judicial documents, as well as scholarly texts such as medical, divinatory, and liturgical works. Additionally, cuneiform tablets feature literary compositions and monumental inscriptions [3].\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eTo create cuneiform signs, scribes made impressions on wet clay tablets using a stylus [4]. Compared to papyrus or leather these tablets are notably thicker. The size and format of the tablets varied according to the time-period, genre, and location. Due to the plasticity of wet clay, not only the script but also seals (see\u003cstrong\u003e\u0026nbsp;Figure 6\u003c/strong\u003e and\u003cstrong\u003e\u0026nbsp;Figure 7\u003c/strong\u003e) and unintentional fingerprints\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003e(see \u003cstrong\u003eFigure 1\u003c/strong\u003e) were impressed on the tablets. Consequently, the genre, form, depth of the wedges, lighting conditions, and shadows all influence the reading of cuneiform texts. Therefore, the most appropriate digitization method for cuneiform tablets is through 3D scanning.\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eCuneiform tablets have been found at numerous sites throughout the Near East [3]. Today, many excavated tablets are stored in various museums worldwide. Some museums have digitized their collections or are in the process of doing so\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003ethrough the Cuneiform Digital Library Initiative (CDLI)\u003ca href=\"#_ftn1\" name=\"_ftnref1\" title=\"\"\u003e\u003c/a\u003e\u003csup\u003e1\u003c/sup\u003e and the electronic Babylonian Library (eBL).\u003ca href=\"#_ftn2\" name=\"_ftnref2\" title=\"\"\u003e\u003c/a\u003e\u003csup\u003e2\u003c/sup\u003e A common preservation method involves creating hand copies of the tablets. Digital preservation methods include digitizing these hand copies and using 2D photography of the tablets. These digital methods are among the most common and cost-efficient approaches to preservation.\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe cuneiform script was adapted to multiple languages, including from the Semitic and Indo-European language families, but also other isolated languages. Of particular interest in Ancient Mesopotamia are Sumerian and Akkadian. Sumerian, an isolated agglutinative language, is attested in texts from the 28th century BCE until the end of the cuneiform era around the transition to the common era [5]. However, around the second half of the 2nd millennium BCE, Sumerian gradually ceased to be a spoken language and transitioned into a scholarly and liturgical language [6]. \u003cstrong\u003e\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe second language of particular interest is Akkadian, a Semitic language originating in northern Mesopotamia. Akkadian is first attested around the 26th century BCE and remained in use until the end of the cuneiform culture, with the aforementioned tablet from 75 CE being written in Akkadian [7]. Although still debated, it is believed that by the mid-5th century BCE, Akkadian began to be replaced by Aramaic as a spoken language, while Akkadian persisted as a written and scholarly language [8]. During this period, the cuneiform script was also gradually replaced by the Aramaic script.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eThe dataset published here contains 3D models of cuneiform tablets from the Slemani Museum, located in the Kurdistan Region of Iraq. This collection hosts more than 6,000 tablets and other cuneiform bearing artefacts (like bricks, prisms, cones, etc.) from various time periods and provenances, with only about 10% of them published.\u003c/p\u003e\n\u003cp\u003eThis initiative aimed to scan as many cuneiform tablets and inscribed objects as possible within a ten-day period. To enable a comprehensive analysis of the forms, layouts, and scripts from different historical periods, we selected tablets and objects spanning various chronological phases for scanning. In total, 92 artifacts featuring cuneiform texts in Sumerian and Akkadian were scanned. The distribution of the types of objects is as follows:\u0026nbsp;\u003c/p\u003e\n\u003ctable border=\"1\" cellspacing=\"0\" cellpadding=\"0\" width=\"184\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 131px;\"\u003e\n \u003cp\u003eCuneiform tablets\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 53px;\"\u003e\n \u003cp\u003e87\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 131px;\"\u003e\n \u003cp\u003eBrick\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 53px;\"\u003e\n \u003cp\u003e1\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 131px;\"\u003e\n \u003cp\u003eClay Hand\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 53px;\"\u003e\n \u003cp\u003e1\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 131px;\"\u003e\n \u003cp\u003eSumerian Stela\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 53px;\"\u003e\n \u003cp\u003e1\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 131px;\"\u003e\n \u003cp\u003ePrisms\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 53px;\"\u003e\n \u003cp\u003e2\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 131px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eTotal\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 53px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e92\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n\u003cp\u003e\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eClassifying them by genre results in the following distribution:\u0026nbsp;\u003c/p\u003e\n\u003ctable border=\"1\" cellspacing=\"0\" cellpadding=\"0\" width=\"159\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 105px;\"\u003e\n \u003cp\u003eArchival Texts\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 54px;\"\u003e\n \u003cp\u003e87\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 105px;\"\u003e\n \u003cp\u003eLiterary Texts\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 54px;\"\u003e\n \u003cp\u003e1\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 105px;\"\u003e\n \u003cp\u003eMonumental\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 54px;\"\u003e\n \u003cp\u003e4\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 105px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eTotal\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 54px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e92\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n\u003cp\u003eArchival texts, the most frequently attested genre in the dataset, include letters, payment receipts, delivery documents, and judicial records. However, some literary and monumental texts are also present.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eDue to publication rights we can only publish in this dataset 44 cuneiform tablets. The distribution of these texts according to genre and subtype is as follow:\u0026nbsp;\u003c/p\u003e\n\u003ctable border=\"1\" cellspacing=\"0\" cellpadding=\"0\" width=\"217\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 156px;\"\u003e\n \u003cp\u003eArchival Texts\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 31px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 31px;\"\u003e\n \u003cp\u003e41\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 156px;\"\u003e\n \u003cul\u003e\n \u003cli\u003eletters\u003c/li\u003e\n \u003c/ul\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 31px;\"\u003e\n \u003cp\u003e18\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 31px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 156px;\"\u003e\n \u003cul\u003e\n \u003cli\u003eadministrative\u003c/li\u003e\n \u003c/ul\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 31px;\"\u003e\n \u003cp\u003e20\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 31px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 156px;\"\u003e\n \u003cul\u003e\n \u003cli\u003elegal\u003c/li\u003e\n \u003c/ul\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 31px;\"\u003e\n \u003cp\u003e3\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 31px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 156px;\"\u003e\n \u003cp\u003eLiterary Texts\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 31px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 31px;\"\u003e\n \u003cp\u003e1\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 156px;\"\u003e\n \u003cp\u003eMonumental\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 31px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 31px;\"\u003e\n \u003cp\u003e2\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 156px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eTotal\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 31px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 31px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e44\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n\u003ch3\u003eComputational Analysis\u003c/h3\u003e\n\u003cp\u003eThis dataset is being integrated into the CuBeDa (Cuneiform Benchmark Dataset) series, which is essential for training neural networks in the development of e.g. 3D cuneiform OCR (Optical Character Recognition). Initial steps in this direction are discussed in [13].\u003c/p\u003e\n\u003cp\u003eFor OCR development, the 3D models can be annotated as shown in [14]. Additionally, 2D images generated from these 3D models can be annotated, as demonstrated in [10]. These images can also be employed for cuneiform sign recognition, as recently shown by Mikulinsky et al. [15] who demonstrate how prototype alignment techniques can enhance recognition accuracy by leveraging structural configurations, particularly beneficial for rare signs. Moreover, this dataset supports the comparative analysis of 3D versus 2D geometric morphometrics of cuneiform signs across different periods, contributing to the identification of traits in digital palaeography.\u003c/p\u003e\n\u003ch3\u003eFingerprints Analysis\u003c/h3\u003e\n\u003cp\u003eThis dataset also has significant potential for fingerprint analysis. Some of the tablets exhibit fingerprints, and their extraction is enhanced by MSII (see Figure 1). Such analysis could provide insights into the age and sex of the scribes, offering a deeper understanding of the people who created these ancient documents. A recent study conducted by Jon Ross et al [16] presented new biometric evidence from fingerprints preserved on ceramic objects from the Late Bronze Age II stratum at Tel Burna, providing insights into the demographics of pottery production. The analysis revealed that female potters predominantly produced ceramics.\u003c/p\u003e\n\u003ch3\u003ePhilological Editions\u003c/h3\u003e\n\u003cp\u003eUntil now, 3D models have not been particularly useful for philological editions. This can be attributed to various factors, such as a lack of interest in the method or the restricted availability to only a select group of scholars. However, as digital methods become more widespread, the use and inclusion of 3D datasets are adding new levels of quality and possibilities. Processing with GigaMesh and the application of MSII further enhance their potential for reuse in classical philological editions. Consequently, these high-resolution 3D models are emerging as a valuable tool for philological analysis, complementing traditional photographs and handmade copies.\u003c/p\u003e\n\u003cp\u003eAs example, Karen Radner [17] published a photograph of a badly damaged prism fragment, SM 043619 (T-7619), and tentatively identified it as part of Esarhaddon Episode 24: Nin. A, VI 60 [18] or Esarhaddon 1 VI 60 in the Oracc (Open Richly Annotated Cuneiform Corpus) edition, based on the clarity of only the penultimate line (see Figure 2).\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eHowever, after processing the 3D model and applying MSII filtering, it became possible to more clearly identify additional lines, leading to a definitive identification as Esarhaddon 1 III 27-45 (see Figures 3, 4, 5) according to the Oracc classification. Although this fragment does not introduce significant variants to the already reconstructed inscription, aside from line breaks, it demonstrates the potential of this method in enhancing textual analysis.\u003c/p\u003e\n\u003ch3\u003eSeals extraction\u003c/h3\u003e\n\u003cp\u003eUsing different coloration after applying the MSII stamp and cylinder seals become more visible (see Figures 6 and 7).\u003c/p\u003e"},{"header":"Method","content":"\u003cp\u003eThe dataset follows a set of criteria for the publication of 3D models in Assyriology, established by Homburg et al. [9] and further applied by Homburg et al. [10] in their work on annotated 3D models of cuneiform tablets, and also Homburg et al. [11] in their Pipeline for Processing the tablet from Haft Tappeh.\u003c/p\u003e\n\n\u003cp\u003eThe 3D acquisition was conducted by Luis S\u0026aacute;enz and Hemin Fatah at the museum, with technical support from Steffen Bauer (Germany). Scanning took place over a ten-day period at the end of September 2023, using a Hexagon SmartScan structured light scanner (SLS) within the museum\u0026rsquo;s facilities.\u003c/p\u003e\n\n\u003cp\u003eFor cuneiform tablets we used the so-called small field of view with a diagonal of 125mm (S-125). With these settings, the scanner achieves a resolution of around 40\u0026micro;m in the plane (X-Y) and a resolution limit of 5\u0026micro;m in height (Z). According to the manufacturer\u0026apos;s specification sheet, the feature accuracy is 9\u0026micro;m. No color information was captured.\u003c/p\u003e\n\n\u003cp\u003eThis configuration is particularly well suited for small tablets, approximately up to 10 cm\u0026sup2; in size, which could be fully covered within the 125 mm field of view and typically required 15\u0026ndash;20 minutes to scan. Each small tablet typically required 6\u0026ndash;8 scan layers. In contrast, larger tablets demanded significantly more time due to the need for multiple overlapping scans. Using lenses with a wider field of view was not a viable solution, as it would have resulted in reduced resolution and, consequently, less accurate 3D models. To improve efficiency, we introduced a measuring procedure using a ruler to define scanning sections of approximately 11 cm\u0026sup2;, ensuring an overlap of about 1 cm between adjacent scans. This method allowed us to reduce the number of redundant scans and streamline the scanning process for larger tablets. All layers were spatially aligned and subsequently optimized before being merged into a single 3D mesh.\u003c/p\u003e\n\u003ch2\u003eData Record\u003c/h2\u003e\n\u003cp\u003eThe 3D acquisition of the tablets was conducted using the aforementioned scanner at the museum\u0026rsquo;s facilities in late September 2023. Immediately after each scan, the resulting 3D model was temporarily stored in the scanner\u0026rsquo;s proprietary format. At the end of each day, the models were exported as PLY files and processed in GigaMesh, following these steps:\u003c/p\u003e\n\n\u003col start=\"1\" type=\"1\"\u003e\n\u003cli\u003eAutomatic Mesh Polishing\u003c/li\u003e\n\u003c/ol\u003e\n\u003cul\u003e\n\u003cli\u003eCleaning (Erosion)\u003c/li\u003e\n\u003cli\u003eFilling (Dilation)\u003c/li\u003e\n\u003c/ul\u003e\n\u003col start=\"2\" type=\"1\"\u003e\n\u003cli\u003eManual orientation to establish the artifact\u0026rsquo;s coordinate system\u003c/li\u003e\n\u003cli\u003eApplication of the Multi-Scale Integral Invariants (MSII) to visualize and extract the smallest details\u003c/li\u003e\n\u003cli\u003eExport to the PLY format\u003c/li\u003e\n\u003c/ol\u003e\n\n\u003cp\u003eFor steps 1 and 2, a video tutorial available on the Heidelberger Dokumentserver and on YouTube via the GigaMesh\u003cu\u003e channel\u003c/u\u003e. A tutorial for step 3 is also available on the same YouTube channel.\u003c/p\u003e\n\n\u003cp\u003eThe primary data product in this publication consists of exported Stanford Polygon (PLY) files. These files are named with the postfix \u0026apos;GMOCF,\u0026apos; indicating that they have been processed with \u003cstrong\u003eG\u003c/strong\u003eiga\u003cstrong\u003eM\u003c/strong\u003eesh, \u003cstrong\u003eo\u003c/strong\u003eriented, \u003cstrong\u003ec\u003c/strong\u003eleaned, and had holes \u003cstrong\u003ef\u003c/strong\u003eilled. Orientation defines the front of the tablet, which in the context of a 3D model may differ from the traditional philological designations of \u0026apos;obverse\u0026apos; and \u0026apos;reverse.\u0026apos; For example, the front in a 3D model might be what is traditionally considered the reverse if the obverse is significantly damaged. The cleaning process involves removing outliers, zero-area vertices, etc, while filling holes, which are typically small missing sections in the 3D surface. These holes, although often not easily visible, can impact the results of the MSII filter. Additionally, the cleaning and filling process ensures that the models are watertight, making them suitable for 3D printing. GigaMesh also enriches the mesh with metadata, particularly indicating which vertices were artificially added during the filling process. Additionally, metadata about the meshes are exported into CSV (see \u0026ldquo;Quality Control\u0026rdquo;). \u003c/p\u003e\n\n\u003cp\u003eSubsequently, six images per 3D model were rendered, depicting the front, back, top, bottom, left, and right sides of the cuneiform tablets. Each view was individually saved as a PNG file. These images were then compiled into a single-page PDF in the fatcross view using LaTeX.\u003c/p\u003e\n\n\u003cp\u003eFollowing this, we applied the MSII filter to visualize fine details, such as fingerprints, damaged cuneiform signs and seal impressions (see Figures 1,5,6 and 7). These details are crucial for further studies (see \u0026ldquo;Reuse Potential\u0026rdquo;).\u003c/p\u003e\n\n\u003cp\u003eThe processed 3D models are available for visualization and download via heidICON, a digital image database maintained by Heidelberg University that hosts a wide range of visual research materials, including photographs, drawings, manuscripts, and other multimedia. Within heidICON, the models are published under the Propylaeum Pool, specifically in the subrepository \u0026ldquo;Keilschrifttafeln aus Sulaymaniyah\u0026rdquo;. The 3D models are enriched with metadata in accordance with the standards of the Heidelberg University Library.\u003c/p\u003e\n\n\u003cp\u003eThe metadata concerning the quality assessments of the models (see section \u0026ldquo;Quality Control\u0026rdquo;) is published in heidDATA, the university\u0026rsquo;s institutional repository for open research data, and is linked to the corresponding entry in the Propylaeum Pool.\u003c/p\u003e\n\n\u003cp\u003eThese repositories follow the principles of the Heidelberg Cuneiform Benchmark Dataset (HeiCuBeDa), a subrepository dedicated to 3D models of cuneiform and related artifacts from the Hilprecht Collection in Jena. HeiCuBeDa, initiated by Hubert Mara, is also hosted on heidDATA [12].\u003c/p\u003e\n\n\u003cp\u003eThis work can therefore be summarized in the following steps: (i) acquisition of archaeological artifacts, (ii) generation of 3D models, (iii) mesh cleaning, (iv) manual orientation, (v) rendering of 2D images for each tablet\u0026rsquo;s side, employing a normalized grayscale and consistent resolution across the entire dataset, (vi) standardized visualization of these images in a PDF using the fatcross rendering method, (vii) application of the 3D MSII filter, and (viii) enrichment of the metadata.\u003c/p\u003e\n\u003ch3\u003eSpatial coverage\u003c/h3\u003e\n\u003cp\u003eThe tablets are stored and were scanned at:\u003c/p\u003e\n\n\u003cp\u003eSulaymaniyah Museum, Sulaymaniyah, Kurdistan Region of Iraq. \u003c/p\u003e\n\n\u003cp\u003eNorthern boundary: 35.5579703\u003c/p\u003e\n\u003cp\u003eSouthern boundary: 35.5570720\u003c/p\u003e\n\u003cp\u003eEastern boundary: 45.42572153\u003c/p\u003e\n\u003cp\u003eWestern boundary: 45.4257376\u003c/p\u003e\n\n\u003cp\u003eThe provenance of most of the tablets is largely unknown. However, some tablets are confirmed to originate from cities such as Sippar, Larsa, Adab, Isin and Ninive. These locations suggest that the tablets likely originate from both northern and southern Mesopotamia:\u003c/p\u003e\n\n\u003cp\u003eNorthern boundary: 36.3667\u0026deg; N, 43.1500\u0026deg; E (Near the ruins of Nineveh, close to modern-day Mosul, Iraq)\u003c/p\u003e\n\u003cp\u003eSouthern boundary: 31.2267\u0026deg; N, 45.8667\u0026deg; E (Near the ancient city of Larsa, close to modern-day Tell as-Senkereh, Iraq)\u003c/p\u003e\n\u003cp\u003eWestern boundary: 33.0608\u0026deg; N, 44.2544\u0026deg; E (Near the ancient city of Sippar, close to modern-day Abu Habbah, Iraq)\u003c/p\u003e\n\u003cp\u003eEastern Boundary: 32.1133\u0026deg; N, 45.5000\u0026deg; E (Near the ancient city of Adab, close to modern-day Bismaya, Iraq)\u003c/p\u003e\n\u003ch3\u003eTemporal coverage\u003c/h3\u003e\n\u003cp\u003eThe total temporal span of the tablets ranges from 2330 to 330 BCE. The distribution of tablets across different historical periods is as follows:\u003c/p\u003e\n\n\u003cul\u003e\n\u003cli\u003eSargonic (2334\u0026ndash;2154 BCE): 7 tablets\u003c/li\u003e\n\u003cli\u003eUr III (2110\u0026ndash;2003 BCE): 16 tablets\u003c/li\u003e\n\u003cli\u003eNeo-Assyrian (1000\u0026ndash;600 BCE): 2 tablets\u003c/li\u003e\n\u003cli\u003eNeo-Babylonian (1025\u0026ndash;330 BCE): 1 tablet\u003c/li\u003e\n\u003c/ul\u003e\n\u003ch3\u003eSampling strategy\u003c/h3\u003e\n\u003cp\u003eTo comprehensively analyze the forms, layouts, and scripts across different periods, we selected tablets from various chronological phases for scanning. This approach allows us to capture a broad spectrum of variations in tablet format and writing styles. By including samples from multiple periods, we aim to identify and compare the distinct characteristics and developments in the tablets\u0026apos; physical and textual features throughout history.\u003c/p\u003e\n\u003ch3\u003eData type\u003c/h3\u003e\n\u003cp\u003eProcessed data (3D meshes, PNG and PDF) and interpretation of data (transliteration, translation and quality assessments in CSV-Format).\u003c/p\u003e\n\u003ch3\u003eFormat names and versions\u003c/h3\u003e\n\u003cp\u003ePolygon File Format (PLY),\u003c/p\u003e\n\u003cp\u003ePortable Document Format (PDF),\u003c/p\u003e\n\u003cp\u003ePortable Network Graphic (PNG),\u003c/p\u003e\n\u003cp\u003eComma-Separated Values (CSV)\u003c/p\u003e\n\u003ch3\u003eCreation dates\u003c/h3\u003e\n\u003cp\u003e23/09/2023 \u0026ndash; 02/10/2023: 3D scans of cuneiform tablets and other archaeological artifacts at the Sulaymaniyah-Museum and generation of the 3D models.\u003c/p\u003e\n\n\u003cp\u003e11/2023: Creation of PLY, PDF and PNG files.\u003c/p\u003e\n\n\u003cp\u003e26/04/2025: Creation of CSV files.\u003c/p\u003e\n\u003ch3\u003eDataset Creators\u003c/h3\u003e\n\u003cp\u003eLuis S\u0026aacute;enz, performed the 3D scans and exported them in PLY Format, processed the models as described in Method the orientation, created the Metadata and the png renderings of the 3D models as well as the PDF rendetrings in fatcross view.\u003c/p\u003e\n\n\u003cp\u003eHemin Nure Fatah, Support and help for creation of the 3D scans at the Museum\u003c/p\u003e\n\n\u003cp\u003eSteffen Bauer, Supported scanner and Scanner Software, Created the 3D renderings and quality assessment metadata of all cuneiform 3D scans. \u003c/p\u003e\n\n\u003cp\u003eNyan Naser Hama Hasan, prepared the tablets for scanning\u003c/p\u003e\n\n\u003cp\u003eHashim Hama Abdulla, Coordinated the scanning process at the Museum.\u003c/p\u003e\n\u003ch3\u003eLanguage\u003c/h3\u003e\n\u003cp\u003eEnglish in the metadata. Akkadian and Sumerian on the scanned tablets.\u003c/p\u003e\n\u003ch3\u003eLicense\u003c/h3\u003e\n\u003cp\u003eCC-BY SA 4.0\u003c/p\u003e\n\u003ch3\u003eRepository location\u003c/h3\u003e\n\u003cp\u003eThe data is published on heidICON, the repository of the University Library of heidelberg, under the following URL https://heidicon.ub.uni-heidelberg.de/pool/cuneiform_sulaimaniya. \u003c/p\u003e\n\n\u003cp\u003eThe cuneiform tablet scans are published there using the following DOI:\u003c/p\u003e\n\n\u003cp\u003eSM.036042: https://doi.org/10.11588/heidicon/23958493\u003c/p\u003e\n\u003cp\u003eSM.036311: https://doi.org/10.11588/heidicon/23958494\u003c/p\u003e\n\u003cp\u003eSM.036359: https://doi.org/10.11588/heidicon/23958495\u003c/p\u003e\n\u003cp\u003eSM.036373: https://doi.org/10.11588/heidicon/23958496 \u003c/p\u003e\n\u003cp\u003eSM.036389: https://doi.org/10.11588/heidicon/23958497\u003c/p\u003e\n\u003cp\u003eSM.036390: https://doi.org/10.11588/heidicon/23958498\u003c/p\u003e\n\u003cp\u003eSM.036413: https://doi.org/10.11588/heidicon/23958499\u003c/p\u003e\n\u003cp\u003eSM.036425: https://doi.org/10.11588/heidicon/23958500\u003c/p\u003e\n\u003cp\u003eSM.036432: https://doi.org/10.11588/heidicon/23958501\u003c/p\u003e\n\u003cp\u003eSM.036475: https://doi.org/10.11588/heidicon/23958502\u003c/p\u003e\n\u003cp\u003eSM.036486: https://doi.org/10.11588/heidicon/23958503\u003c/p\u003e\n\u003cp\u003eSM.036497: https://doi.org/10.11588/heidicon/23958504\u003c/p\u003e\n\u003cp\u003eSM.036547: https://doi.org/10.11588/heidicon/23958505\u003c/p\u003e\n\u003cp\u003eSM.037023: https://doi.org/10.11588/heidicon/23958506\u003c/p\u003e\n\u003cp\u003eSM.037244: https://doi.org/10.11588/heidicon/23958507\u003c/p\u003e\n\u003cp\u003eSM.037315: https://doi.org/10.11588/heidicon/23958508\u003c/p\u003e\n\u003cp\u003eSM.037316: https://doi.org/10.11588/heidicon/23958509\u003c/p\u003e\n\u003cp\u003eSM.037317: https://doi.org/10.11588/heidicon/23958510\u003c/p\u003e\n\u003cp\u003eSM.037318: https://doi.org/10.11588/heidicon/23958511\u003c/p\u003e\n\u003cp\u003eSM.037319: https://doi.org/10.11588/heidicon/23958512\u003c/p\u003e\n\u003cp\u003eSM.037321: https://doi.org/10.11588/heidicon/23958513\u003c/p\u003e\n\u003cp\u003eSM.037322: https://doi.org/10.11588/heidicon/23958514\u003c/p\u003e\n\u003cp\u003eSM.037323: https://doi.org/10.11588/heidicon/23958515\u003c/p\u003e\n\u003cp\u003eSM.037324: https://doi.org/10.11588/heidicon/23958516\u003c/p\u003e\n\u003cp\u003eSM.037326: https://doi.org/10.11588/heidicon/23958517\u003c/p\u003e\n\u003cp\u003eSM.037327: https://doi.org/10.11588/heidicon/23958518\u003c/p\u003e\n\u003cp\u003eSM.037377: https://doi.org/10.11588/heidicon/23958519\u003c/p\u003e\n\u003cp\u003eSM.037447: https://doi.org/10.11588/heidicon/23958520\u003c/p\u003e\n\u003cp\u003eSM.037798: https://doi.org/10.11588/heidicon/23958521\u003c/p\u003e\n\u003cp\u003eSM.037942: https://doi.org/10.11588/heidicon/23958522\u003c/p\u003e\n\u003cp\u003eSM.039043: https://doi.org/10.11588/heidicon/23958523\u003c/p\u003e\n\u003cp\u003eSM.039055: https://doi.org/10.11588/heidicon/23958524\u003c/p\u003e\n\u003cp\u003eSM.039085: https://doi.org/10.11588/heidicon/23958525\u003c/p\u003e\n\u003cp\u003eSM.039156: https://doi.org/10.11588/heidicon/23958526\u003c/p\u003e\n\u003cp\u003eSM.039816: https://doi.org/10.11588/heidicon/23958527\u003c/p\u003e\n\u003cp\u003eSM.039833: https://doi.org/10.11588/heidicon/23958528\u003c/p\u003e\n\u003cp\u003eSM.039886: https://doi.org/10.11588/heidicon/23958529\u003c/p\u003e\n\u003cp\u003eSM.039941: https://doi.org/10.11588/heidicon/23958530\u003c/p\u003e\n\u003cp\u003eSM.039977: https://doi.org/10.11588/heidicon/23958531\u003c/p\u003e\n\u003cp\u003eSM.041015: https://doi.org/10.11588/heidicon/23958532\u003c/p\u003e\n\u003cp\u003eSM.041016: https://doi.org/10.11588/heidicon/23958533\u003c/p\u003e\n\u003cp\u003eSM.041017: https://doi.org/10.11588/heidicon/23958534\u003c/p\u003e\n\u003cp\u003eSM.043618: https://doi.org/10.11588/heidicon/23958535\u003c/p\u003e\n\u003cp\u003eSM.043619: https://doi.org/10.11588/heidicon/23958536\u003c/p\u003e\n\n\u003cp\u003eThe CSV tables are there using the following DOI: https://doi.org/10.11588/DATA/4FROK0.\u003c/p\u003e\n\n\u003ch3\u003ePublication date\u003c/h3\u003e\n\u003cp\u003eThe dataset in heidICON was published on heidICON on 27/02/2025, the dataset in heiDATA on 09/05/2025.\u003c/p\u003e\n\u003ch3\u003eConstraints\u003c/h3\u003e\n\u003cp\u003eOne of the primary constraints encountered in this initiative is the reluctance of museums to grant publication rights for the 3D models. As a result, we are limited to publishing 3D models of tablets that have already been made publicly available in previous publications. This restriction significantly impacts the breadth of material we can share.\u003c/p\u003e\n\u003ch2\u003eTechnical Validation\u003c/h2\u003e\n\u003cp\u003eThe 3D models generated in this study are of high quality, offering both accuracy and reliability for subsequent analysis. To ensure and document these standards, GigaMesh was employed not only for mesh processing but also for exporting comprehensive technical metadata (see \u0026ldquo;Repository location\u0026rdquo;). \u003c/p\u003e\n\u003cp\u003eDespite the small physical dimensions of the tablets, the high-resolution scans result in a substantial number of vertices\u0026mdash;ranging from 1.1 million to 10.6 million per model, with an average of approximately 2.7 million. A key metric indicating the completeness of the 3D scans is the proportion of synthetic vertices introduced during post-processing in GigaMesh. This proportion remains low, averaging just 1.8%, with a maximum of about 4%. Notably, nearly two-thirds of all models contain fewer than 2% synthetic vertices.\u003c/p\u003e\n\u003cp\u003eOccasional artifacts, such as wave-like distortions in localized scan segments, may occur due to the structured light scanning method, particularly with the specific scanner employed (cf. models SM 039043, SM 037244). These should not be mistaken for intentional imprints such as fingerprints or handprints (see Figure 1).\u003c/p\u003e\n\u003cp\u003eAnother rare but observable artifact involves the exaggeration of geometric edges caused by strong color contrasts, especially where inventory numbers are written in black ink on the clay tablets (e.g., SM 037318, SM 036425).\u003c/p\u003e\n\u003cp\u003eImportantly, these infrequent artifacts do not compromise the overall quality or usability of the 3D models for further research and analytical applications.\u003c/p\u003e"},{"header":"Declarations","content":"\u003ch2\u003eData Availability\u003c/h2\u003e\n\u003cp\u003eThe data is published on heidICON, the repository of the University Library of heidelberg, under the following URL https://heidicon.ub.uni-heidelberg.de/pool/cuneiform_sulaimaniya.\u003c/p\u003e\n\u003cp\u003eThe CSV tables are there using the following DOI: https://doi.org/10.11588/DATA/4FROK0.\u003c/p\u003e\n\u003ch2\u003eAcknowledgements\u003c/h2\u003e\n\u003cp\u003eWe would like to express our gratitude to the Sulaymaniyah Museum for their invaluable support and collaboration.\u003c/p\u003e\n\u003cp\u003eWe also extend our sincere thanks to Timo Homburg for developing a parsing script that converted the metadata from CSV tables into JSON files, the required format for repository submission.\u003c/p\u003e\n\u003ch2\u003eAuthor Contributions\u003c/h2\u003e\n\u003cp\u003eL.S., M.A., and Sh.G. conceived the study, M.A. and Sh.G. supervised and funded the project. L.S. and S.B. were in charge of data creation and curation. S.B. and H.M. were responsible for the methodology, and H.M. created the GigaMesh software used for 3D model analysis. L.S. prepared the first draft of the manuscript with all figures, as well as the initial analysis of 3D models. All authors participated in writing, editing, and revising the manuscript.\u003c/p\u003e\n\u003ch2\u003eFunding Statement\u003c/h2\u003e\n\u003cp\u003eThis initiative has been funded by the Paleography Powered by Machine Learning: Analyzing Cuneiform Documents. LMU-TAU Cooperation Fund, internal grant, and by the Babylonian Industrialism in the Age of Empires. Internal Ariel University grant funding RA2300000111.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n\u003cli\u003eKrebernik, M. Zur Struktur und Geschichte des \u0026auml;lteren sumerischen Onomastikons, in \u003cem\u003eAlter Orient und Altes Testament\u003c/em\u003e \u003cstrong\u003e296\u003c/strong\u003e 1-74 (M\u0026uuml;nster Ugarit-Verlag 2002).\u003c/li\u003e\n\u003cli\u003eSachs, A. J. The latest datable cuneiform tablets, in \u003cem\u003eAlter Orient und Altes Testament\u003c/em\u003e \u003cstrong\u003e25\u003c/strong\u003e, 379-398 (Kevelaer Neukirchener Verlag 1976)\u003c/li\u003e\n\u003cli\u003eStreck, M. P. Gro\u0026szlig;es Fach Altorientalistik: Der Umfang des keilschriftlichen Textkorpus, in \u003cem\u003eMitteilungen der Deutschen Orient-Gesellschaft\u003c/em\u003e \u003cstrong\u003e142\u003c/strong\u003e, 35-58, (2010).\u003c/li\u003e\n\u003cli\u003eGlassner, J.-J. \u003cem\u003eThe invention of cuneiform: writing in Sumer\u003c/em\u003e (Baltimore Johns Hopkins University Press 2003).\u003c/li\u003e\n\u003cli\u003eSallaberger, W. Das Ende des Sumerischen: Tod und Nachleben einer altmesopotamischen Sprache, in \u003cem\u003eM\u0026uuml;nchner Forschungen zur historischen Sprachwissenschaft\u003c/em\u003e \u003cstrong\u003e2\u003c/strong\u003e, 108\u0026ndash;140 (Bremen Hempen Verlag 2004).\u003c/li\u003e\n\u003cli\u003eMaul, S. M. K\u0026uuml;chensumerisch oder hohe Kunst der Exegese? \u0026Uuml;berlegungen zur Bewertung akkadischer Interlinear\u0026uuml;bersetzungen von Emesal-Texten, in \u003cem\u003eAlter Orient und Altes Testament\u003c/em\u003e \u003cstrong\u003e247\u003c/strong\u003e, 253-267 (Neukirchen-Vluyn Neukirchener Verlag 1997).\u003c/li\u003e\n\u003cli\u003eStreck, M. P. Akkadian and Cuneiform in \u003cem\u003eHandbuch der Orientalistik\u003c/em\u003e \u003cstrong\u003e152\u003c/strong\u003e, 66-74 (Leiden, Boston Brill 2021).\u003c/li\u003e\n\u003cli\u003eHackl, J. The Death of Akkadian as a Written and Spoken Language, in \u003cem\u003eHandbuch der Orientalistik\u003c/em\u003e \u003cstrong\u003e152\u003c/strong\u003e, 1459-1477 (Leiden, Boston Brill 2021).\u003c/li\u003e\n\u003cli\u003eHomburg, T., Cramer, A. , Raddatz, L. \u0026amp; Mara, H. Metadata schema and ontology for capturing and processing of 3D cultural heritage objects https://www.nature.com/articles/s40494-021-00561-w (2021).\u003c/li\u003e\n\u003cli\u003eHomburg, T. Zwick, R., Mara, H. \u0026amp; Bruhn, K.-C. Annotated 3D-Models of Cuneiform Tablets https://openarchaeologydata.metajnl.com/articles/10.5334/joad.92 (2022).\u003c/li\u003e\n\u003cli\u003eHomburg, T., Zwick, R., Bruhn, K.-C. \u0026amp; H. Mara 3D Data Derivatives of the Haft Tappeh Processing Pipeline https://cdli.mpiwg-berlin.mpg.de/articles/cdlj/2022-1 (2022). \u003c/li\u003e\n\u003cli\u003eMara, H. \u0026amp; Bogacz, B. Breaking the Code on Broken Tablets: The Learning Challenge for Annotated Cuneiform Script in Normalized 2D and 3D Datasets https://ieeexplore.ieee.org/document/8978050 (2019).\u003c/li\u003e\n\u003cli\u003eSt\u0026ouml;tzner, E., Homburg, T. \u0026amp; Mara, H. CNN based Cuneiform Sign Detection Learned from Annotated 3D Renderings and Mapped Photographs with Illumination Augmentation https://arxiv.org/abs/2308.11277 (2023).\u003c/li\u003e\n\u003cli\u003eSt\u0026ouml;tzner, E., Homburg, T. , Bullenkamp, J. P. \u0026amp; Mara, H. R-CNN based PolygonalWedge Detection Learned from Annotated 3D Renderings and Mapped Photographs of Open Data Cuneiform Tablets https://diglib.eg.org/items/e9e8eb87-cb45-4c57-b2bc-44fb347b8c4d (2023).\u003c/li\u003e\n\u003cli\u003eMikulinsky, R., Alper, M., Gordin, S., Jim\u0026eacute;nez, E., Cohen, Y., and Averbuch-Elor, H. ProtoSnap: Prototype alignment for cuneiform signs in https://openreview.net/forum?id=XHTirKsQV6 (2024).\u003c/li\u003e\n\u003cli\u003eRoss, J., Fowler, K. D., and Shai, I. New Fingerprint Evidence for Female Potters in Late Bronze Age Canaan: The Demographics of Potters and Division of Labour at Tel Burna doi: https://www.sciencedirect.com/science/article/abs/pii/S0278416523000491?via%3Dihub (2023).\u003c/li\u003e\n\u003cli\u003eRadner, K. Cuneiform Inscriptions in the Archaeological Museum of Sulaimaniya \u003cem\u003eAltorientalische Forschungen\u003c/em\u003e \u003cstrong\u003e52\u003c/strong\u003e, 98-103 (2011).\u003c/li\u003e\n\u003cli\u003eBorger, R. \u003cem\u003eDie Inschriften Asarhaddons K\u0026ouml;nigs von Assyrien\u003c/em\u003e (Graz Biblio-Verlag 1956).\u003c/li\u003e\n\u003c/ol\u003e"},{"header":"Footnotes","content":"\u003col\u003e\n \u003cli\u003ehttps://cdli.earth.\u003c/li\u003e\n \u003cli\u003ehttps://www.ebl.lmu.de.\u003c/li\u003e\n\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":true,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true},"keywords":"","lastPublishedDoi":"10.21203/rs.3.rs-8661845/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-8661845/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eThis paper presents high-resolution 3D models of cuneiform tablets from the Sulaymaniyah Museum, digitized using the Hexagon SmartScan 3D with approximately 10\u0026micro;m spatial resolution. The dataset includes 44 publishable tablets spanning from 2330\u0026thinsp;\u0026minus;\u0026thinsp;330 BCE, processed with GigaMesh open-source software. The 3D models enable advanced research applications including OCR development, fingerprint analysis, improved philological editions, and seal extraction. Enhanced visualization using Multi-Scale Integral Invariants (MSII) filtering reveals previously unreadable text and fine details. This initiative contributes significantly to digital palaeography and cultural heritage preservation while following established criteria for 3D data publication in Assyriology.\u003c/p\u003e","manuscriptTitle":"3D Dataset of Cuneiform Bearing Objects at the Sulaymaniyah Museum","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2026-03-18 17:00:00","doi":"10.21203/rs.3.rs-8661845/v1","editorialEvents":[{"type":"communityComments","content":0}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"12f946c9-bf60-4d9c-a03d-0f1d7d7ca9f2","owner":[],"postedDate":"March 18th, 2026","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[],"tags":[],"updatedAt":"2026-03-18T17:00:00+00:00","versionOfRecord":[],"versionCreatedAt":"2026-03-18 17:00:00","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-8661845","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-8661845","identity":"rs-8661845","version":["v1"]},"buildId":"XKTyCvWXoU3ODBz1xrDgd","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}