July 13, 2026 Dataset Open Access

Radoynova, Martina; Pantze, Samuel; De, Trina; Günther, Ulrik; Yakimovich, Artur

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-4" xsi:schemaLocation="http://datacite.org/schema/kernel-4 http://schema.datacite.org/meta/kernel-4.1/metadata.xsd">
  <identifier identifierType="DOI">10.14278/rodare.4780</identifier>
  <creators>
    <creator>
      <creatorName>Radoynova, Martina</creatorName>
      <givenName>Martina</givenName>
      <familyName>Radoynova</familyName>
      <nameIdentifier nameIdentifierScheme="ORCID" schemeURI="http://orcid.org/">0009-0003-4358-0391</nameIdentifier>
      <affiliation>Center for Advanced Systems Understanding (CASUS), Görlitz, Germany</affiliation>
    </creator>
    <creator>
      <creatorName>Pantze, Samuel</creatorName>
      <givenName>Samuel</givenName>
      <familyName>Pantze</familyName>
      <nameIdentifier nameIdentifierScheme="ORCID" schemeURI="http://orcid.org/">0009-0000-1388-8959</nameIdentifier>
      <affiliation>Center for Advanced Systems Understanding (CASUS), Görlitz, Germany</affiliation>
    </creator>
    <creator>
      <creatorName>De, Trina</creatorName>
      <givenName>Trina</givenName>
      <familyName>De</familyName>
      <nameIdentifier nameIdentifierScheme="ORCID" schemeURI="http://orcid.org/">0000-0003-1111-9851</nameIdentifier>
      <affiliation>Center for Advanced Systems Understanding (CASUS), Görlitz, Germany</affiliation>
    </creator>
    <creator>
      <creatorName>Günther, Ulrik</creatorName>
      <givenName>Ulrik</givenName>
      <familyName>Günther</familyName>
      <nameIdentifier nameIdentifierScheme="ORCID" schemeURI="http://orcid.org/">0000-0002-1179-8228</nameIdentifier>
      <affiliation>Center for Advanced Systems Understanding (CASUS), Görlitz, Germany</affiliation>
    </creator>
    <creator>
      <creatorName>Yakimovich, Artur</creatorName>
      <givenName>Artur</givenName>
      <familyName>Yakimovich</familyName>
      <nameIdentifier nameIdentifierScheme="ORCID" schemeURI="http://orcid.org/">0000-0003-2458-4904</nameIdentifier>
      <affiliation>Center for Advanced Systems Understanding (CASUS), Görlitz, Germany</affiliation>
    </creator>
  </creators>
  <titles>
    <title>Metric-Guided Synthetic Image Data Rendering for Deep Learning compatible with Agentic AI</title>
  </titles>
  <publisher>Rodare</publisher>
  <publicationYear>2026</publicationYear>
  <subjects>
    <subject>deep learning</subject>
    <subject>synthetic data</subject>
    <subject>procedural image rendering</subject>
    <subject>Blender</subject>
    <subject>computer vision</subject>
    <subject>instance segmentation</subject>
  </subjects>
  <dates>
    <date dateType="Issued">2026-07-13</date>
  </dates>
  <language>en</language>
  <resourceType resourceTypeGeneral="Dataset"/>
  <alternateIdentifiers>
    <alternateIdentifier alternateIdentifierType="url">https://rodare.hzdr.de/record/4780</alternateIdentifier>
  </alternateIdentifiers>
  <relatedIdentifiers>
    <relatedIdentifier relatedIdentifierType="URL" relationType="IsIdenticalTo">https://www.hzdr.de/publications/Publ-43634</relatedIdentifier>
    <relatedIdentifier relatedIdentifierType="DOI" relationType="IsVersionOf">10.14278/rodare.4779</relatedIdentifier>
    <relatedIdentifier relatedIdentifierType="URL" relationType="IsPartOf">https://rodare.hzdr.de/communities/health</relatedIdentifier>
    <relatedIdentifier relatedIdentifierType="URL" relationType="IsPartOf">https://rodare.hzdr.de/communities/rodare</relatedIdentifier>
  </relatedIdentifiers>
  <version>Version 1</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;This repository contains the datasets, 3D rendering scene files, and trained model weights accompanying the manuscript &lt;strong&gt;&amp;quot;Metric-Guided Synthetic Image Data Rendering for Deep Learning compatible with Agentic AI.&amp;quot;&lt;/strong&gt; The deposited files provide the complete procedural pipeline and resulting artefacts used to evaluate zero-shot instance segmentation of virological plaque assays (based on the &lt;a href="https://rodare.hzdr.de/record/3003"&gt;VACVPlaque dataset&lt;/a&gt;).&lt;/p&gt;

&lt;p&gt;Repository Contents&lt;/p&gt;

&lt;ul&gt;
	&lt;li&gt;
	&lt;p&gt;&lt;strong&gt;&lt;code&gt;datasets/&lt;/code&gt;&lt;/strong&gt; Contains the synthetic image datasets generated at three varying levels of perceived realism (High, Medium, Low) and in two sizes (Small [S]: 100 images; Large [L]: 1,000 images). It also includes the &lt;code&gt;Mix&lt;/code&gt; dataset, which consists of 90 high-realism synthetic images and 10 real images.&lt;/p&gt;
	&lt;/li&gt;
	&lt;li&gt;
	&lt;p&gt;&lt;strong&gt;&lt;code&gt;blender/&lt;/code&gt;&lt;/strong&gt; Includes the procedural 3D scene files used to generate the synthetic datasets via Blender. These files can be used manually or coupled with the SynthClaw agentic skill for automated rendering.&lt;/p&gt;

	&lt;ul&gt;
		&lt;li&gt;
		&lt;p&gt;&lt;code&gt;Scene_High.blend&lt;/code&gt;&lt;/p&gt;
		&lt;/li&gt;
		&lt;li&gt;
		&lt;p&gt;&lt;code&gt;Scene_Medium.blend&lt;/code&gt;&lt;/p&gt;
		&lt;/li&gt;
		&lt;li&gt;
		&lt;p&gt;&lt;code&gt;Scene_Low.blend&lt;/code&gt;&lt;/p&gt;
		&lt;/li&gt;
	&lt;/ul&gt;
	&lt;/li&gt;
	&lt;li&gt;
	&lt;p&gt;&lt;strong&gt;&lt;code&gt;model_weights/&lt;/code&gt;&lt;/strong&gt; Contains the trained single-shot architecture model weights and threshold configurations used to evaluate the zero-shot performance of each generated dataset.&lt;/p&gt;

	&lt;ul&gt;
		&lt;li&gt;
		&lt;p&gt;&lt;strong&gt;Evaluated Conditions:&lt;/strong&gt; &lt;code&gt;High_L&lt;/code&gt;, &lt;code&gt;High_S&lt;/code&gt;, &lt;code&gt;Medium_L&lt;/code&gt;, &lt;code&gt;Medium_S&lt;/code&gt;, &lt;code&gt;Low_L&lt;/code&gt;, &lt;code&gt;Low_S&lt;/code&gt;, and &lt;code&gt;Mix&lt;/code&gt;.&lt;/p&gt;
		&lt;/li&gt;
		&lt;li&gt;
		&lt;p&gt;&lt;strong&gt;Files per Condition:&lt;/strong&gt; Each subdirectory includes the best and last model weights (&lt;code&gt;weights_best.h5&lt;/code&gt;, &lt;code&gt;weights_last.h5&lt;/code&gt;) alongside the corresponding threshold optimization parameters (&lt;code&gt;thresholds1.json&lt;/code&gt;, &lt;code&gt;thresholds2.json&lt;/code&gt;) for large and small object segmentation.&lt;/p&gt;
		&lt;/li&gt;
	&lt;/ul&gt;
	&lt;/li&gt;
&lt;/ul&gt;</description>
  </descriptions>
</resource>