<?xml version="1.0" encoding="UTF-8"?>
<rss xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:rdf="http://www.w3.org/1999/02/22-rdf-syntax-ns#" xmlns:taxo="http://purl.org/rss/1.0/modules/taxonomy/" version="2.0">
  <channel>
    <title>topic Deep Learning Detecting Swimming Pools in ArcGIS Image Analyst Questions</title>
    <link>https://community.esri.com/t5/arcgis-image-analyst-questions/deep-learning-detecting-swimming-pools/m-p/268075#M240</link>
    <description>&lt;HTML&gt;&lt;HEAD&gt;&lt;/HEAD&gt;&lt;BODY&gt;&lt;P&gt;I have been working through my own version of detecting swimming pools.&lt;/P&gt;&lt;P&gt;I have use the same imagery (albeit, a subset) used by ESRI.&lt;/P&gt;&lt;P&gt;I think I have taken reasonable training sites.&amp;nbsp; I have tried to be meticulous with selecting training sites.&amp;nbsp; The ones I choose seem similar to those found in the ESRI examples.&lt;/P&gt;&lt;P&gt;&lt;/P&gt;&lt;P&gt;I have used the same (and different parameters for training the model and detecting objects).&amp;nbsp; I have taken the defaults, added values shown in ESRI examples, and tried some of my own options for parameters such as:&lt;/P&gt;&lt;P&gt;&lt;/P&gt;&lt;P&gt;image tile, stride size&lt;/P&gt;&lt;P&gt;grid, scaling, ratio&lt;/P&gt;&lt;P&gt;padding, threshold, etc&lt;/P&gt;&lt;P&gt;epochs&lt;/P&gt;&lt;P&gt;&lt;/P&gt;&lt;P&gt;The "learning curve" has shown concave to "jagged" (which I believe means, overfitted).&amp;nbsp; Like I mentioned, I have tried this scenario numerous times with the different combinations of parameters for both training and detecting.&lt;/P&gt;&lt;P&gt;&lt;/P&gt;&lt;P&gt;I am no where close to successfully detecting swimming pools.&amp;nbsp; I mostly get features other than pools.&lt;/P&gt;&lt;P&gt;I am attaching my PPKX with the training and detected objects (and the image), about 23 Mb.&amp;nbsp; I do have a model zip file, but it is approx. ~200 Mb, so it may be too large.&amp;nbsp; I am attaching the EMD file and the PNG that shows the loss curve.&amp;nbsp; This is one is jagged.&amp;nbsp; 25 epochs.&lt;/P&gt;&lt;P&gt;&lt;/P&gt;&lt;P&gt;The attachment has about 15 sample sites, but I have done this same process using a larger image and approx. 80 samples that I think are "good" samples of swimming pools...same kind of results.&lt;/P&gt;&lt;P&gt;&lt;/P&gt;&lt;P&gt;If anyone can provide some suggestions or guidance, I would like to successfully attempt to find swimming pools....and then work with detecting other kinds of features.&lt;/P&gt;&lt;P&gt;&lt;/P&gt;&lt;P&gt;I am using an XPS Dell business machine.&amp;nbsp; Pro 2.5.&amp;nbsp; I have the deep learning framework installed.&amp;nbsp; No GPU or NVIDIA card.&amp;nbsp; I can successfully run the full process, but the results are very bad.&lt;/P&gt;&lt;P&gt;&lt;/P&gt;&lt;P&gt;I have viewed and use: &amp;nbsp;&lt;A class="link-titled" href="https://hub.arcgis.com/datasets/e03a6691f20c4e898e2bd084c3e96c8a" title="https://hub.arcgis.com/datasets/e03a6691f20c4e898e2bd084c3e96c8a"&gt;Deep Learning: Detect swimming pools&lt;/A&gt;&amp;nbsp; as a guide in addition to the ESRI help and other ArcGIS.learn notebooks and other information provided by ESRI.&lt;/P&gt;&lt;P&gt;&lt;/P&gt;&lt;P&gt;Thank you for any insight or let me know if I should be contacting someone else at ESRI.&amp;nbsp; I would also like to know if there is a process to go through to logically attempt different parameter values that can improve both training and detecting results.&lt;/P&gt;&lt;P&gt;&lt;/P&gt;&lt;P&gt;Nate&lt;/P&gt;&lt;/BODY&gt;&lt;/HTML&gt;</description>
    <pubDate>Sat, 18 Apr 2020 00:02:20 GMT</pubDate>
    <dc:creator>NathanJennings</dc:creator>
    <dc:date>2020-04-18T00:02:20Z</dc:date>
    <item>
      <title>Deep Learning Detecting Swimming Pools</title>
      <link>https://community.esri.com/t5/arcgis-image-analyst-questions/deep-learning-detecting-swimming-pools/m-p/268075#M240</link>
      <description>&lt;HTML&gt;&lt;HEAD&gt;&lt;/HEAD&gt;&lt;BODY&gt;&lt;P&gt;I have been working through my own version of detecting swimming pools.&lt;/P&gt;&lt;P&gt;I have use the same imagery (albeit, a subset) used by ESRI.&lt;/P&gt;&lt;P&gt;I think I have taken reasonable training sites.&amp;nbsp; I have tried to be meticulous with selecting training sites.&amp;nbsp; The ones I choose seem similar to those found in the ESRI examples.&lt;/P&gt;&lt;P&gt;&lt;/P&gt;&lt;P&gt;I have used the same (and different parameters for training the model and detecting objects).&amp;nbsp; I have taken the defaults, added values shown in ESRI examples, and tried some of my own options for parameters such as:&lt;/P&gt;&lt;P&gt;&lt;/P&gt;&lt;P&gt;image tile, stride size&lt;/P&gt;&lt;P&gt;grid, scaling, ratio&lt;/P&gt;&lt;P&gt;padding, threshold, etc&lt;/P&gt;&lt;P&gt;epochs&lt;/P&gt;&lt;P&gt;&lt;/P&gt;&lt;P&gt;The "learning curve" has shown concave to "jagged" (which I believe means, overfitted).&amp;nbsp; Like I mentioned, I have tried this scenario numerous times with the different combinations of parameters for both training and detecting.&lt;/P&gt;&lt;P&gt;&lt;/P&gt;&lt;P&gt;I am no where close to successfully detecting swimming pools.&amp;nbsp; I mostly get features other than pools.&lt;/P&gt;&lt;P&gt;I am attaching my PPKX with the training and detected objects (and the image), about 23 Mb.&amp;nbsp; I do have a model zip file, but it is approx. ~200 Mb, so it may be too large.&amp;nbsp; I am attaching the EMD file and the PNG that shows the loss curve.&amp;nbsp; This is one is jagged.&amp;nbsp; 25 epochs.&lt;/P&gt;&lt;P&gt;&lt;/P&gt;&lt;P&gt;The attachment has about 15 sample sites, but I have done this same process using a larger image and approx. 80 samples that I think are "good" samples of swimming pools...same kind of results.&lt;/P&gt;&lt;P&gt;&lt;/P&gt;&lt;P&gt;If anyone can provide some suggestions or guidance, I would like to successfully attempt to find swimming pools....and then work with detecting other kinds of features.&lt;/P&gt;&lt;P&gt;&lt;/P&gt;&lt;P&gt;I am using an XPS Dell business machine.&amp;nbsp; Pro 2.5.&amp;nbsp; I have the deep learning framework installed.&amp;nbsp; No GPU or NVIDIA card.&amp;nbsp; I can successfully run the full process, but the results are very bad.&lt;/P&gt;&lt;P&gt;&lt;/P&gt;&lt;P&gt;I have viewed and use: &amp;nbsp;&lt;A class="link-titled" href="https://hub.arcgis.com/datasets/e03a6691f20c4e898e2bd084c3e96c8a" title="https://hub.arcgis.com/datasets/e03a6691f20c4e898e2bd084c3e96c8a"&gt;Deep Learning: Detect swimming pools&lt;/A&gt;&amp;nbsp; as a guide in addition to the ESRI help and other ArcGIS.learn notebooks and other information provided by ESRI.&lt;/P&gt;&lt;P&gt;&lt;/P&gt;&lt;P&gt;Thank you for any insight or let me know if I should be contacting someone else at ESRI.&amp;nbsp; I would also like to know if there is a process to go through to logically attempt different parameter values that can improve both training and detecting results.&lt;/P&gt;&lt;P&gt;&lt;/P&gt;&lt;P&gt;Nate&lt;/P&gt;&lt;/BODY&gt;&lt;/HTML&gt;</description>
      <pubDate>Sat, 18 Apr 2020 00:02:20 GMT</pubDate>
      <guid>https://community.esri.com/t5/arcgis-image-analyst-questions/deep-learning-detecting-swimming-pools/m-p/268075#M240</guid>
      <dc:creator>NathanJennings</dc:creator>
      <dc:date>2020-04-18T00:02:20Z</dc:date>
    </item>
  </channel>
</rss>

