June 30, 2023 Dataset Open Access

Hussein Mohammed

DataCite XML Export

<?xml version='1.0' encoding='utf-8'?>
<resource xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xmlns="http://datacite.org/schema/kernel-3" xsi:schemaLocation="http://datacite.org/schema/kernel-3 http://schema.datacite.org/meta/kernel-3/metadata.xsd">
  <identifier identifierType="DOI">10.25592/uhhfdm.12671</identifier>
  <creators>
    <creator>
      <creatorName>Hussein Mohammed</creatorName>
      <nameIdentifier nameIdentifierScheme="ORCID" schemeURI="http://orcid.org/">0000-0001-5020-3592</nameIdentifier>
      <affiliation>Universität Hamburg</affiliation>
    </creator>
  </creators>
  <titles>
    <title>Mini-dataset for VL-Models fine-tuning (VL-Tune-dataset-mini)</title>
  </titles>
  <publisher>Universität Hamburg</publisher>
  <publicationYear>2023</publicationYear>
  <subjects>
    <subject>Vision-Language Models, Dataset</subject>
  </subjects>
  <dates>
    <date dateType="Issued">2023-06-30</date>
  </dates>
  <language>en</language>
  <resourceType resourceTypeGeneral="Dataset"/>
  <alternateIdentifiers>
    <alternateIdentifier alternateIdentifierType="url">https://www.fdr.uni-hamburg.de/record/12671</alternateIdentifier>
  </alternateIdentifiers>
  <relatedIdentifiers>
    <relatedIdentifier relatedIdentifierType="DOI" relationType="IsPartOf">10.25592/uhhfdm.12670</relatedIdentifier>
  </relatedIdentifiers>
  <version>1.0</version>
  <rightsList>
    <rights rightsURI="https://creativecommons.org/licenses/by/4.0/legalcode">Creative Commons Attribution 4.0 International</rights>
    <rights rightsURI="info:eu-repo/semantics/openAccess">Open Access</rights>
  </rightsList>
  <descriptions>
    <description descriptionType="Abstract">&lt;p&gt;A minimal dataset of 125 image-text pairs&amp;nbsp;and 10&amp;nbsp;text queries&amp;nbsp;for fine-tuning vision-language models on manuscript images.&amp;nbsp;It is dedicated to the task of text-based image retrieval, and splited into &amp;quot;train&amp;quot; and &amp;quot;test&amp;quot; sets. The train set&amp;nbsp;consists of 100 image-text pairs, while the test set consists of 25 image-text pairs. This dataset is constructed from the following sources:&lt;/p&gt;

&lt;p&gt;- images from&amp;nbsp;the &lt;a href="http://spotting.univ-rouen.fr/"&gt;DocExplore&lt;/a&gt; dataset&amp;nbsp;of medieval manuscripts.&lt;/p&gt;

&lt;p&gt;- images from&amp;nbsp;two manuscripts from Al-Ḥarīrī,&amp;nbsp;&lt;em&gt;Maqāmāt&lt;/em&gt;,&amp;nbsp;&amp;copy; Paris,&amp;nbsp;Biblioth&amp;egrave;que nationale de France. D&amp;eacute;partement des manuscrits,&amp;nbsp;namely MS arabe 3929 and MS arabe 5847.&lt;/p&gt;

&lt;p&gt;- the descriptions in the text files are prepared by&amp;nbsp;Martina Dinelli&lt;/p&gt;

&lt;p&gt;The research for this work was funded by the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation) under Germany&amp;#39;s Excellence Strategy &amp;ndash; EXC 2176 &amp;lsquo;Understanding Written Artefacts: Material, Interaction and Transmission in Manuscript Cultures&amp;#39;, project no. 390893796. The research was conducted within the scope of the Centre for the Study of Manuscript Cultures (CSMC) at Universit&amp;auml;t Hamburg.&lt;/p&gt;</description>
  </descriptions>
</resource>