November 14, 2025 Dataset Open Access

Chekhonin, Paul; Korten, Till; Gerçek, Alinda Ezgi; Hassan, Maleeha; Steinbach, Peter

Citation Style Language JSON Export

{
  "title": "Training Data and Models for the paper: Data-efficient U-Net for Segmentation of Carbide Microstructures in SEM Images of Steel Alloys", 
  "issued": {
    "date-parts": [
      [
        2025, 
        11, 
        14
      ]
    ]
  }, 
  "id": "4124", 
  "publisher": "Rodare", 
  "author": [
    {
      "family": "Chekhonin, Paul"
    }, 
    {
      "family": "Korten, Till"
    }, 
    {
      "family": "Ger\u00e7ek, Alinda Ezgi"
    }, 
    {
      "family": "Hassan, Maleeha"
    }, 
    {
      "family": "Steinbach, Peter"
    }
  ], 
  "type": "dataset", 
  "DOI": "10.14278/rodare.4124", 
  "abstract": "<p>This dataset contains scanning electron microscopy (SEM) images of steel alloys, including paired secondary electron (SE2) and in-lens (InLens) channels, with corresponding binary segmentation labels. The data supports full reproduction of results presented in the referenced manuscript.</p>\n\n<p>&nbsp;</p>\n\n<p><strong>Dataset Description</strong></p>\n\n<ul>\n\t<li><strong>Content:</strong>&nbsp;13 pairs of SEM images of two reactor pressure vessel (RPV) steels:\n\n\t<ul>\n\t\t<li><em>JFL</em>: IAEA reference RPV base metal steel</li>\n\t\t<li><em>ANP-10</em>: Western type RPV steel</li>\n\t</ul>\n\t</li>\n\t<li><strong>Acquisition:</strong>\n\t<ul>\n\t\t<li><em>JFL</em>: Zeiss NVision 40 microscope</li>\n\t\t<li><em>ANP-10</em>: Zeiss Ultra 55 microscope</li>\n\t\t<li>Both SE and InLens detectors used simultaneously.</li>\n\t</ul>\n\t</li>\n\t<li><strong>Resolution:</strong>&nbsp;2048 &times; 1404 pixels per image\n\t<ul>\n\t\t<li>2048 px width corresponds to 14.3 &micro;m (JFL) or 11.5 &micro;m (ANP-10).</li>\n\t</ul>\n\t</li>\n</ul>\n\n<p>Using the dataset to reproduce the results of the manuscript</p>\n\n<p>Download the zip file into the&nbsp;<code>data/</code> subdirectory of the code repository and extract the archive:</p>\n\n<pre><code class=\"language-bash\">cd data/\nunzip data.zip</code></pre>\n\n<p><strong>Dataset Structure</strong></p>\n\n<p>These directories contain the relevant data for the manuscript:</p>\n\n<p><code>cloud/</code><br>\n<code>\u251c-\u2500 preprocessed/</code><br>\n<code>\u2502 &nbsp; \u251c\u2500\u2500 hold-out/</code><br>\n<code>\u2502 &nbsp; \u251c\u2500\u2500 images/</code><br>\n<code>\u2502 &nbsp; \u2514\u2500\u2500 labels/</code><br>\n<code>\u251c\u2500\u2500 processed_tiles/</code><br>\n<code>\u2502 &nbsp; \u251c\u2500\u2500 images/</code><br>\n<code>\u2502 &nbsp; \u2514\u2500\u2500 labels/</code><br>\n<code>\u251c\u2500\u2500 tb_logs/</code><br>\n<code>\u2502 &nbsp; \u251c\u2500\u2500 unet_model/</code></p>\n\n<p><strong>Preprocessed</strong></p>\n\n<p>pre-processed whole images and corresponding labels</p>\n\n<p><strong>Processed Tiles</strong></p>\n\n<p>tiled images and labels</p>\n\n<p><strong>tb_logs</strong></p>\n\n<p>trained model weights</p>"
}