{"id":1152172,"date":"2021-03-02T01:00:52","date_gmt":"2021-03-02T09:00:52","guid":{"rendered":"https:\/\/www.esri.com\/arcgis-blog\/?post_type=blog&#038;p=1152172"},"modified":"2021-10-07T09:27:18","modified_gmt":"2021-10-07T16:27:18","slug":"deep-learning-with-arcgis-pro-tips-tricks-part-2","status":"publish","type":"blog","link":"https:\/\/www.esri.com\/arcgis-blog\/products\/arcgis-pro\/imagery\/deep-learning-with-arcgis-pro-tips-tricks-part-2","title":{"rendered":"Deep Learning with ArcGIS Pro Tips &amp; Tricks: Part 2"},"author":19291,"featured_media":0,"parent":0,"menu_order":0,"comment_status":"open","ping_status":"closed","template":"","format":"standard","meta":{"_acf_changed":false,"_searchwp_excluded":""},"categories":[22931,22941],"tags":[42181,186132,757081,665211,115262],"industry":[],"product":[36581,36561],"class_list":["post-1152172","blog","type-blog","status-publish","format-standard","hentry","category-imagery","category-mapping","tag-arcgis-pro","tag-deep-learning","tag-deep-learning-models","tag-geoai","tag-imagery","product-arcgis-living-atlas","product-arcgis-pro"],"acf":{"related_articles":[{"ID":1071621,"post_author":"19291","post_date":"2023-11-15 18:05:31","post_date_gmt":"2023-11-16 02:05:31","post_content":"","post_title":"Deep Learning with ArcGIS Pro Tips &amp; Tricks: Part 1","post_excerpt":"","post_status":"publish","comment_status":"open","ping_status":"closed","post_password":"","post_name":"deep-learning-with-arcgis-pro-tips-tricks","to_ping":"","pinged":"","post_modified":"2023-11-15 09:05:39","post_modified_gmt":"2023-11-15 17:05:39","post_content_filtered":"","post_parent":0,"guid":"https:\/\/www.esri.com\/arcgis-blog\/?post_type=blog&#038;p=1071621","menu_order":0,"post_type":"blog","post_mime_type":"","comment_count":"1","filter":"raw"},{"ID":1147822,"post_author":"8452","post_date":"2021-02-23 19:27:02","post_date_gmt":"2021-02-24 03:27:02","post_content":"","post_title":"Pretrained deep learning models update (February 2021)","post_excerpt":"","post_status":"publish","comment_status":"open","ping_status":"closed","post_password":"","post_name":"announcing-new-pretrained-models-at-fedgis","to_ping":"","pinged":"","post_modified":"2021-11-19 08:58:54","post_modified_gmt":"2021-11-19 16:58:54","post_content_filtered":"","post_parent":0,"guid":"https:\/\/www.esri.com\/arcgis-blog\/?post_type=blog&#038;p=1147822","menu_order":0,"post_type":"blog","post_mime_type":"","comment_count":"0","filter":"raw"},{"ID":1035891,"post_author":"6911","post_date":"2020-10-13 11:49:30","post_date_gmt":"2020-10-13 18:49:30","post_content":"","post_title":"Introducing pretrained geospatial deep learning models","post_excerpt":"","post_status":"publish","comment_status":"open","ping_status":"closed","post_password":"","post_name":"introducing-ready-to-use-deep-learning-models","to_ping":"","pinged":"","post_modified":"2021-11-19 09:09:52","post_modified_gmt":"2021-11-19 17:09:52","post_content_filtered":"","post_parent":0,"guid":"https:\/\/www.esri.com\/arcgis-blog\/?post_type=blog&#038;p=1035891","menu_order":0,"post_type":"blog","post_mime_type":"","comment_count":"0","filter":"raw"}],"short_description":"Evaluate your imagery for deep learning and get started using out-of-the-box models such as building footprint extraction. ","flexible_content":[{"acf_fc_layout":"sidebar","content":"<p><span class=\"TextRun SCXW132804305 BCX0\" lang=\"EN-US\" xml:lang=\"EN-US\" data-contrast=\"auto\"><span class=\"NormalTextRun SCXW132804305 BCX0\">In this post we will provide tips and tricks to use Esri&#8217;s pre-trained deep learning models, starting by ensuring your imagery is suitable for deep learning. Then we will break down the input parameters for the Detect Objects using Deep Learning geoprocessing tool, so you can make smart adjustments and optimize the model run for your data and processing environment.<\/span><\/span><\/p>\n","image_reference":false,"layout":"standard","image_reference_figure":"","snippet":"","spotlight_name":"","section_title":"","position":"Center","spotlight_image":false},{"acf_fc_layout":"content","content":"<p>Part 1 of this blog series (<a href=\"https:\/\/www.esri.com\/arcgis-blog\/products\/arcgis-pro\/imagery\/deep-learning-with-arcgis-pro-tips-tricks\/\">Deep Learning with ArcGIS Pro Tips &amp; Tricks: Part 1<\/a>) covered how to set up your environment to perform deep learning. At this point you should have a CUDA-capable GPU, a recent version of ArcGIS Pro, and the deep learning framework for ArcGIS Pro installed. If we were implementing deep learning from scratch, the next step would be to draw hundreds to thousands of training samples. But with Esri&#8217;s pre-trained, publicly available deep learning models, we can start identifying features in our imagery with one tool in ArcGIS Pro. There are out-of-the-box models available for building footprint extraction, road extraction, land cover classification, detecting human settlements, and more. The available models can be viewed and downloaded from the ArcGIS Living Atlas by anyone with an ArcGIS Online subscription.<\/p>\n<p><span data-contrast=\"auto\">To use the inferencing tools in ArcGIS Pro, we will work through the following checklist:<\/span><span data-ccp-props=\"{&quot;201341983&quot;:0,&quot;335559739&quot;:160,&quot;335559740&quot;:259}\">\u00a0<\/span><\/p>\n<ul>\n<li data-leveltext=\"\uf0b7\" data-font=\"Symbol\" data-listid=\"4\" data-aria-posinset=\"1\" data-aria-level=\"1\"><span data-contrast=\"auto\">Download a deep learning model package (dlpk)<\/span><span data-ccp-props=\"{&quot;134233279&quot;:true,&quot;201341983&quot;:0,&quot;335559739&quot;:160,&quot;335559740&quot;:259}\">\u00a0<\/span><\/li>\n<li data-leveltext=\"\uf0b7\" data-font=\"Symbol\" data-listid=\"4\" data-aria-posinset=\"2\" data-aria-level=\"1\"><span data-contrast=\"auto\">Confirm your imagery is suitable for deep\u00a0learning<\/span><span data-ccp-props=\"{&quot;134233279&quot;:true,&quot;201341983&quot;:0,&quot;335559739&quot;:160,&quot;335559740&quot;:259}\">\u00a0<\/span><\/li>\n<li data-leveltext=\"\uf0b7\" data-font=\"Symbol\" data-listid=\"4\" data-aria-posinset=\"3\" data-aria-level=\"1\"><span data-contrast=\"auto\">Determine the resolution of your\u00a0imagery<\/span><span data-ccp-props=\"{&quot;134233279&quot;:true,&quot;201341983&quot;:0,&quot;335559739&quot;:160,&quot;335559740&quot;:259}\">\u00a0<\/span><\/li>\n<li data-leveltext=\"\uf0b7\" data-font=\"Symbol\" data-listid=\"4\" data-aria-posinset=\"4\" data-aria-level=\"1\"><span data-contrast=\"auto\">Review input parameters for the Detect Objects using Deep Learning geoprocessing\u00a0tool<\/span><span data-ccp-props=\"{&quot;134233279&quot;:true,&quot;201341983&quot;:0,&quot;335559739&quot;:160,&quot;335559740&quot;:259}\">\u00a0<\/span><\/li>\n<li data-leveltext=\"\uf0b7\" data-font=\"Symbol\" data-listid=\"4\" data-aria-posinset=\"5\" data-aria-level=\"1\"><span data-contrast=\"auto\">Run the Detect Objects using Deep Learning geoprocessing\u00a0tool<\/span><span data-ccp-props=\"{&quot;134233279&quot;:true,&quot;201341983&quot;:0,&quot;335559739&quot;:160,&quot;335559740&quot;:259}\">\u00a0<\/span><\/li>\n<\/ul>\n<h2>Key Definitions<\/h2>\n<p>.emd: \u00a0Esri Model Definition<\/p>\n<p>TensorFlow: TensorFlow is an end-to-end open source platform for machine learning. It has a comprehensive, flexible ecosystem of tools, libraries and community resources that lets researchers push the state-of-the-art in ML and developers easily build and deploy ML powered applications. (tensorflow.org)<\/p>\n<p>Keras: Keras is an open-source software library that provides a Python interface for artificial neural networks. Keras acts as an interface for the TensorFlow library.<\/p>\n<p>.pb: TensorFlow deep learning model file extension<\/p>\n<p>.h5: Keras deep learning model file extension<\/p>\n<p>.py: Python script<\/p>\n<p>GeoAi: Geographical Artificial Intelligence<\/p>\n<p>Dlpk: Esri Deep learning package<\/p>\n<p>CNN: Convolutional neural networks<\/p>\n<p>Mask R-CNN: Mask RCNN is a deep neural network aimed to solve instance segmentation problem in machine learning or computer vision. In other words, it can separate different objects in a image or a video. You give it a image, it gives you the object bounding boxes, classes and masks. (<a href=\"Simple%20Understanding%20of%20Mask%20RCNN\"><em>Simple Understanding of Mask RCNN<\/em><\/a><em>, Xiang Zhang Apr 22, 2018)<\/em><\/p>\n<p>Epoch: The number of epochs is a hyperparameter that defines the number times that the learning algorithm will work through the entire training dataset (<em><a href=\"https:\/\/machinelearningmastery.com\/difference-between-a-batch-and-an-epoch\/#:~:text=at%20an%20epoch.-,What%20Is%20an%20Epoch%3F,update%20the%20internal%20model%20parameters.\">Difference Between a Batch and an Epoch in a Neural Network<\/a>, Jason Brownlee, July 20, 2018)<\/em><\/p>\n<h2>Download deep learning model package<\/h2>\n<p>A deep learning model package (.dlpk) contains the files and data required to run deep learning inferencing tools for object detection or image classification. The package can be uploaded to your portal as a DLPK item and used as the input to deep learning raster analysis tools.<\/p>\n<p>Deep learning model packages must contain an\u00a0<a href=\"https:\/\/enterprise.arcgis.com\/en\/portal\/latest\/use\/deep-learning-in-raster-analysis.htm#ESRI_SECTION1_9E58B8C77BC44859ABF286623E19BE8D\">Esri model definition file<\/a>\u00a0(.emd) and a trained model file. The trained model file extension depends on the framework you used to train the model. For example, if you trained your model using TensorFlow, the model file will be a\u00a0.pb<strong>\u00a0<\/strong>file, while a model trained using Keras will generate an .h5 file. Depending on the model framework and options you used to train your model, you may need to include a Python raster function (.py) or additional files. You can include multiple trained model files in a single deep learning model package. (Note that a forthcoming blog post in this series will cover how to train your own model).<\/p>\n<p><a href=\"https:\/\/pro.arcgis.com\/en\/pro-app\/2.7\/get-started\/releases-and-patches.htm#ESRI_SECTION2_99AF6DD6246C4E1AA7F2ED22D092260E\">Most packages<\/a>\u00a0can be opened in any version of\u00a0ArcGIS Pro. By default, the contents of a package are stored in the\u00a0&lt;User Documents&gt;\\ArcGIS\\Packages\u00a0folder. You can change this location in the\u00a0<a href=\"https:\/\/pro.arcgis.com\/en\/pro-app\/2.7\/get-started\/share-and-download-options.htm\">Share and download options<\/a>. Functionality in the package that is not supported at the version of\u00a0ArcGIS Pro\u00a0being used to consume the package is not available.<\/p>\n<p>The easiest way to find Esri&#8217;s out-of-the box models is through the catalog pane in ArcGIS Pro:<\/p>\n<ul>\n<li>Open a new or an existing ArcGIS Pro Project<\/li>\n<li>Navigate to the Catalog Pane<\/li>\n<li>Click Portal<\/li>\n<li>Select Living Atlas Icon<\/li>\n<li>Type dlpk in the search bar &amp; hit enter<\/li>\n<\/ul>\n<p>A list of publicly available dlpks will show in your search results. To view, download or review the properties of a .dlpk, or to add or remove files from your .dlpk, right-click the <strong>.dlpk<\/strong>\u00a0in the\u00a0<strong>Catalog<\/strong>\u00a0pane and click\u00a0<strong>Properties<\/strong>.<\/p>\n"},{"acf_fc_layout":"image","image":{"ID":1152222,"id":1152222,"title":"Deep Learning Package Property ArcGIS Pro","filename":"Edit-DLPKV1.png","filesize":10832,"url":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2021\/03\/Edit-DLPKV1.png","link":"https:\/\/www.esri.com\/arcgis-blog\/products\/arcgis-pro\/imagery\/deep-learning-with-arcgis-pro-tips-tricks-part-2\/edit-dlpkv1","alt":"Deep Learning Package Property ArcGIS Pro","author":"19291","description":"","caption":"","name":"edit-dlpkv1","status":"inherit","uploaded_to":1152172,"date":"2021-03-01 10:27:13","modified":"2021-03-01 10:27:55","menu_order":0,"mime_type":"image\/png","type":"image","subtype":"png","icon":"https:\/\/www.esri.com\/arcgis-blog\/wp-includes\/images\/media\/default.png","width":558,"height":522,"sizes":{"thumbnail":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2021\/03\/Edit-DLPKV1-213x200.png","thumbnail-width":213,"thumbnail-height":200,"medium":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2021\/03\/Edit-DLPKV1.png","medium-width":279,"medium-height":261,"medium_large":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2021\/03\/Edit-DLPKV1.png","medium_large-width":558,"medium_large-height":522,"large":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2021\/03\/Edit-DLPKV1.png","large-width":558,"large-height":522,"1536x1536":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2021\/03\/Edit-DLPKV1.png","1536x1536-width":558,"1536x1536-height":522,"2048x2048":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2021\/03\/Edit-DLPKV1.png","2048x2048-width":558,"2048x2048-height":522,"card_image":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2021\/03\/Edit-DLPKV1-497x465.png","card_image-width":497,"card_image-height":465,"wide_image":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2021\/03\/Edit-DLPKV1.png","wide_image-width":558,"wide_image-height":522}},"image_position":"center","orientation":"horizontal","hyperlink":""},{"acf_fc_layout":"content","content":"<p>Properties include the following information:<\/p>\n<ul>\n<li><strong>Input<\/strong>\u2014The\u00a0.emd\u00a0file, trained model file, and any additional files that may be required to run the inferencing tools.<\/li>\n<li><strong>Framework<\/strong>\u2014The deep learning framework used to train the model.<\/li>\n<li><strong>ModelConfiguration<\/strong>\u2014The type of model training performed (object detection, pixel classification, or feature classification).<\/li>\n<li><strong>Description<\/strong>\u2014A description of the model. This is optional and editable.<\/li>\n<li><strong>Summary<\/strong>\u2014A brief summary of the model. This is optional and editable.<\/li>\n<li><strong>Tags<\/strong>\u2014Any tags used to identify the package. This is useful for\u00a0.dlpk\u00a0package items stored on your portal.<\/li>\n<\/ul>\n"},{"acf_fc_layout":"image","image":{"ID":1152192,"id":1152192,"title":"DLPK.Properties","filename":"DLPK.Properties.png","filesize":11245,"url":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2021\/03\/DLPK.Properties.png","link":"https:\/\/www.esri.com\/arcgis-blog\/products\/arcgis-pro\/imagery\/deep-learning-with-arcgis-pro-tips-tricks-part-2\/dlpk-properties","alt":"","author":"19291","description":"","caption":"","name":"dlpk-properties","status":"inherit","uploaded_to":1152172,"date":"2021-03-01 10:22:01","modified":"2021-03-01 10:22:01","menu_order":0,"mime_type":"image\/png","type":"image","subtype":"png","icon":"https:\/\/www.esri.com\/arcgis-blog\/wp-includes\/images\/media\/default.png","width":758,"height":535,"sizes":{"thumbnail":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2021\/03\/DLPK.Properties-213x200.png","thumbnail-width":213,"thumbnail-height":200,"medium":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2021\/03\/DLPK.Properties.png","medium-width":370,"medium-height":261,"medium_large":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2021\/03\/DLPK.Properties.png","medium_large-width":758,"medium_large-height":535,"large":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2021\/03\/DLPK.Properties.png","large-width":758,"large-height":535,"1536x1536":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2021\/03\/DLPK.Properties.png","1536x1536-width":758,"1536x1536-height":535,"2048x2048":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2021\/03\/DLPK.Properties.png","2048x2048-width":758,"2048x2048-height":535,"card_image":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2021\/03\/DLPK.Properties-659x465.png","card_image-width":659,"card_image-height":465,"wide_image":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2021\/03\/DLPK.Properties.png","wide_image-width":758,"wide_image-height":535}},"image_position":"center","orientation":"horizontal","hyperlink":""},{"acf_fc_layout":"content","content":"<p>Any property that is edited in the\u00a0<strong>Properties<\/strong>\u00a0window is updated when you click\u00a0<strong>OK<\/strong>. If the\u00a0.<strong>dlpk<\/strong>\u00a0item is being accessed from your portal in the\u00a0<strong>Catalog<\/strong>\u00a0pane, the portal item is updated.<\/p>\n<p>Note that Esri provides out-of-the-box .dlpks that you can use in your deep learning workflow. They can be found under the dlpk section in <a href=\"https:\/\/www.arcgis.com\/home\/search.html?q=owner%3A%22esri_analytics%22%20deep%20learning%20package&amp;t=content&amp;restrict=false\">Esri\u2019s Living Atlas<\/a>.<\/p>\n<h2>Confirm your imagery is suitable for deep learning<\/h2>\n<p><span class=\"NormalTextRun BCX0 SCXW176469006\">To get the best result from an inferencing workflow, you need to first visually inspect your imagery. Deep learning models are optimized to detect features that you can see with your eyes. In this part of the blog, we will focus on a hot topic in\u00a0<\/span><span class=\"NormalTextRun SpellingErrorV2 BCX0 SCXW176469006\">GeoAI<\/span><span class=\"NormalTextRun BCX0 SCXW176469006\">: building detection. We will be covering the required imagery resolution for Esri&#8217;s out-of-the-box<\/span>\u00a0<a href=\"https:\/\/www.arcgis.com\/home\/item.html?id=a6857359a1cd44839781a4f113cd5934\">Building Footprint Extraction \u2013 USA<\/a> deep learning package, <span class=\"TextRun BCX0 SCXW176469006\" lang=\"EN-US\" xml:lang=\"EN-US\" data-contrast=\"auto\"><span class=\"NormalTextRun BCX0 SCXW176469006\">but these tips are relevant to any deep learning model.\u00a0<\/span><\/span><\/p>\n<p><span class=\"TextRun SCXW149479809 BCX0\" lang=\"EN-US\" xml:lang=\"EN-US\" data-contrast=\"auto\"><span class=\"NormalTextRun SCXW149479809 BCX0\">First, confirm you\u00a0<\/span><\/span><span class=\"TextRun SCXW149479809 BCX0\" lang=\"EN-US\" xml:lang=\"EN-US\" data-contrast=\"auto\"><span class=\"NormalTextRun AdvancedProofingIssueV2 SCXW149479809 BCX0\">are able to<\/span><\/span><span class=\"TextRun SCXW149479809 BCX0\" lang=\"EN-US\" xml:lang=\"EN-US\" data-contrast=\"auto\"><span class=\"NormalTextRun SCXW149479809 BCX0\">\u00a0visually locate buildings in your raw imagery. As an example, the image below shows some buildings in Cyprus.<\/span><\/span><span class=\"EOP SCXW149479809 BCX0\" data-ccp-props=\"{&quot;201341983&quot;:0,&quot;335559739&quot;:160,&quot;335559740&quot;:259}\">\u00a0<\/span><\/p>\n"},{"acf_fc_layout":"image","image":{"ID":1152232,"id":1152232,"title":"Cyprus Buildings","filename":"Cyprus-Buildings.png","filesize":569398,"url":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2021\/03\/Cyprus-Buildings.png","link":"https:\/\/www.esri.com\/arcgis-blog\/products\/arcgis-pro\/imagery\/deep-learning-with-arcgis-pro-tips-tricks-part-2\/cyprus-buildings","alt":"","author":"19291","description":"","caption":"","name":"cyprus-buildings","status":"inherit","uploaded_to":1152172,"date":"2021-03-01 10:33:51","modified":"2021-03-01 10:34:06","menu_order":0,"mime_type":"image\/png","type":"image","subtype":"png","icon":"https:\/\/www.esri.com\/arcgis-blog\/wp-includes\/images\/media\/default.png","width":1343,"height":492,"sizes":{"thumbnail":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2021\/03\/Cyprus-Buildings-213x200.png","thumbnail-width":213,"thumbnail-height":200,"medium":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2021\/03\/Cyprus-Buildings.png","medium-width":464,"medium-height":170,"medium_large":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2021\/03\/Cyprus-Buildings.png","medium_large-width":768,"medium_large-height":281,"large":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2021\/03\/Cyprus-Buildings.png","large-width":1343,"large-height":492,"1536x1536":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2021\/03\/Cyprus-Buildings.png","1536x1536-width":1343,"1536x1536-height":492,"2048x2048":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2021\/03\/Cyprus-Buildings.png","2048x2048-width":1343,"2048x2048-height":492,"card_image":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2021\/03\/Cyprus-Buildings-826x303.png","card_image-width":826,"card_image-height":303,"wide_image":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2021\/03\/Cyprus-Buildings.png","wide_image-width":1343,"wide_image-height":492}},"image_position":"center","orientation":"horizontal","hyperlink":""},{"acf_fc_layout":"content","content":"<p>We are able to distinguish the individual rooftops from each other and the surroundings, so this image is a good candidate for building footprint extraction.<\/p>\n<h2>Determine the resolution of your imagery<\/h2>\n<p><span class=\"TextRun SCXW236590656 BCX0\" lang=\"EN-US\" xml:lang=\"EN-US\" data-contrast=\"auto\"><span class=\"NormalTextRun SCXW236590656 BCX0\">The\u00a0<\/span><\/span><a class=\"Hyperlink SCXW236590656 BCX0\" href=\"https:\/\/www.arcgis.com\/home\/item.html?id=a6857359a1cd44839781a4f113cd5934\" target=\"_blank\" rel=\"noreferrer noopener\"><span class=\"TextRun Underlined SCXW236590656 BCX0\" lang=\"EN-US\" xml:lang=\"EN-US\" data-contrast=\"none\"><span class=\"NormalTextRun SCXW236590656 BCX0\" data-ccp-charstyle=\"Hyperlink\">Building Footprint Extraction \u2013 USA<\/span><\/span><\/a><span class=\"TextRun SCXW236590656 BCX0\" lang=\"EN-US\" xml:lang=\"EN-US\" data-contrast=\"auto\"><span class=\"NormalTextRun SCXW236590656 BCX0\"> deep learning package is designed to work with high-resolution images (10\u201340 cm). Other dlpks have different recommended resolutions &#8211; check the dlpk&#8217;s item details page for more information. To determine your imagery&#8217;s resolution:<\/span><\/span><span class=\"EOP SCXW236590656 BCX0\" data-ccp-props=\"{&quot;201341983&quot;:0,&quot;335559739&quot;:160,&quot;335559740&quot;:259}\">\u00a0<\/span><\/p>\n<ul>\n<li>In your ArcGIS Pro project, insert a map if you haven\u2019t already done so.<\/li>\n<li>Drag and drop your image onto the map frame.<\/li>\n<li>Select your image in the table of content, then browse to the image properties.<\/li>\n<li>Browse to the <strong>Source <\/strong>tab and expand <strong>Raster Information<\/strong>.<\/li>\n<li>Locate <strong>Cell Size X<\/strong> and <strong>Cell Size Y<\/strong> in the <strong>Raster Information<\/strong><\/li>\n<li>If the cell size is between 0.1 and 0.4 m, you can proceed with the Building Footprint Extraction dlpk.<\/li>\n<li>If not, you might need to follow one of the below workflows:\n<ul>\n<li>If your cell size is much larger than 0.5 m, you may need to acquire new imagery with higher resolution (remember that larger cell size = lower resolution).<\/li>\n<li>If your cell size is smaller than 0.1 m, resample it to the 0.1-0.5m range.<\/li>\n<li>Or train your own model based on your imagery resolution. More info can be found under <a href=\"https:\/\/pro.arcgis.com\/en\/pro-app\/latest\/tool-reference\/image-analyst\/train-deep-learning-model.htm\">Train Deep Learning Model.<\/a><\/li>\n<\/ul>\n<\/li>\n<\/ul>\n"},{"acf_fc_layout":"image","image":{"ID":1152342,"id":1152342,"title":"Imagery Resolution","filename":"image_resolution.png","filesize":18076,"url":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2021\/03\/image_resolution.png","link":"https:\/\/www.esri.com\/arcgis-blog\/products\/arcgis-pro\/imagery\/deep-learning-with-arcgis-pro-tips-tricks-part-2\/image_resolution","alt":"Checking Image Resolution","author":"19291","description":"","caption":"","name":"image_resolution","status":"inherit","uploaded_to":1152172,"date":"2021-03-01 15:24:32","modified":"2021-03-01 15:25:00","menu_order":0,"mime_type":"image\/png","type":"image","subtype":"png","icon":"https:\/\/www.esri.com\/arcgis-blog\/wp-includes\/images\/media\/default.png","width":613,"height":364,"sizes":{"thumbnail":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2021\/03\/image_resolution-213x200.png","thumbnail-width":213,"thumbnail-height":200,"medium":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2021\/03\/image_resolution.png","medium-width":440,"medium-height":261,"medium_large":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2021\/03\/image_resolution.png","medium_large-width":613,"medium_large-height":364,"large":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2021\/03\/image_resolution.png","large-width":613,"large-height":364,"1536x1536":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2021\/03\/image_resolution.png","1536x1536-width":613,"1536x1536-height":364,"2048x2048":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2021\/03\/image_resolution.png","2048x2048-width":613,"2048x2048-height":364,"card_image":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2021\/03\/image_resolution.png","card_image-width":613,"card_image-height":364,"wide_image":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2021\/03\/image_resolution.png","wide_image-width":613,"wide_image-height":364}},"image_position":"center","orientation":"horizontal","hyperlink":""},{"acf_fc_layout":"content","content":"<p>Follow the <a href=\"https:\/\/www.arcgis.com\/home\/item.html?id=a6857359a1cd44839781a4f113cd5934\">Building Footprint Extraction \u2013 USA<\/a> link to download the package. Esri provides a variety of other deep learning packages that can be found under <a href=\"https:\/\/livingatlas.arcgis.com\/en\/browse\/#d=3&amp;q=type%3A%20deep%20learning%20package&amp;type=tool&#096;\">ArcGIS Living Atlas<\/a> of the World. Note that the downloaded model uses the <a href=\"https:\/\/developers.arcgis.com\/python\/guide\/how-maskrcnn-works\/\">Mask R-CNN<\/a> model architecture implemented using ArcGIS API for Python.<\/p>\n<h2>Understand parameters for inferencing<\/h2>\n<p>With all the prerequisites covered, lets dive into the Detect Objects Using Deep Learning geoprocessing tool parameters. A key element of this process is understanding the different parameters that come with the tool:<\/p>\n<ul>\n<li>Padding<\/li>\n<li>Batch size<\/li>\n<li>Threshold<\/li>\n<li>Return bbox<\/li>\n<\/ul>\n<p>Understanding these parameters will allow you to make smart adjustments and get the most accurate output possible.<\/p>\n<p>Note that no deep learning approach will give you 100 percent accurate results, but adjusting your model parameters and iterating through the process can optimize the accuracy of your model. Below we will discuss the importance of each parameter <span class=\"TextRun SCXW251077548 BCX0\" lang=\"EN-US\" xml:lang=\"EN-US\" data-contrast=\"auto\"><span class=\"NormalTextRun SCXW251077548 BCX0\">a<\/span><\/span><span class=\"TextRun SCXW251077548 BCX0\" lang=\"EN-US\" xml:lang=\"EN-US\" data-contrast=\"auto\"><span class=\"NormalTextRun SCXW251077548 BCX0\">nd how<\/span><\/span><span class=\"TextRun SCXW251077548 BCX0\" lang=\"EN-US\" xml:lang=\"EN-US\" data-contrast=\"auto\"><span class=\"NormalTextRun SCXW251077548 BCX0\">\u00a0to\u00a0<\/span><\/span><span class=\"TextRun SCXW251077548 BCX0\" lang=\"EN-US\" xml:lang=\"EN-US\" data-contrast=\"auto\"><span class=\"NormalTextRun SCXW251077548 BCX0\">adjust the inputs based on your imagery and environment<\/span><\/span><span class=\"TextRun SCXW251077548 BCX0\" lang=\"EN-US\" xml:lang=\"EN-US\" data-contrast=\"auto\"><span class=\"NormalTextRun SCXW251077548 BCX0\">.<\/span><\/span><span class=\"EOP SCXW251077548 BCX0\" data-ccp-props=\"{&quot;201341983&quot;:0,&quot;335559739&quot;:160,&quot;335559740&quot;:259}\">\u00a0<\/span><\/p>\n<h3>Padding<\/h3>\n<p>The first parameter is the padding of the model. Padding is the border area from which the model will discard detections, as they tend to be of truncated buildings that span multiple tiles during inferencing. We stride over the padded region, so buildings that are discarded because they lie at the edge in one pass of the model inferencing, are detected in the second pass of the inferencing when they lie at the center of the tiles due to this striding. This means that with the padding parameter being adjusted, the model will adjust the stride of each tile as it runs the inferencing workflow. For example, if we introduce a padding of 32 px (pixels) on a model that is inferencing 128 px tiles, the model will stride the tile by 32 px inside the 4 edges of the tile. If the centroid of the detected feature is within the padded tile, it will pass as a building in this example.<\/p>\n"},{"acf_fc_layout":"image","image":{"ID":1152352,"id":1152352,"title":"Padding V3","filename":"Padding-V3.gif","filesize":659071,"url":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2021\/03\/Padding-V3.gif","link":"https:\/\/www.esri.com\/arcgis-blog\/products\/arcgis-pro\/imagery\/deep-learning-with-arcgis-pro-tips-tricks-part-2\/padding-v3","alt":"","author":"19291","description":"","caption":"","name":"padding-v3","status":"inherit","uploaded_to":1152172,"date":"2021-03-01 15:27:20","modified":"2021-03-01 15:27:20","menu_order":0,"mime_type":"image\/gif","type":"image","subtype":"gif","icon":"https:\/\/www.esri.com\/arcgis-blog\/wp-includes\/images\/media\/default.png","width":853,"height":480,"sizes":{"thumbnail":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2021\/03\/Padding-V3-213x200.gif","thumbnail-width":213,"thumbnail-height":200,"medium":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2021\/03\/Padding-V3.gif","medium-width":464,"medium-height":261,"medium_large":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2021\/03\/Padding-V3.gif","medium_large-width":768,"medium_large-height":432,"large":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2021\/03\/Padding-V3.gif","large-width":853,"large-height":480,"1536x1536":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2021\/03\/Padding-V3.gif","1536x1536-width":853,"1536x1536-height":480,"2048x2048":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2021\/03\/Padding-V3.gif","2048x2048-width":853,"2048x2048-height":480,"card_image":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2021\/03\/Padding-V3-826x465.gif","card_image-width":826,"card_image-height":465,"wide_image":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2021\/03\/Padding-V3.gif","wide_image-width":853,"wide_image-height":480}},"image_position":"center","orientation":"horizontal","hyperlink":""},{"acf_fc_layout":"content","content":"<p><span class=\"TextRun SCXW54123816 BCX0\" lang=\"EN-US\" xml:lang=\"EN-US\" data-contrast=\"auto\"><span class=\"NormalTextRun SCXW54123816 BCX0\">I<\/span><\/span><span class=\"TextRun SCXW54123816 BCX0\" lang=\"EN-US\" xml:lang=\"EN-US\" data-contrast=\"auto\"><span class=\"NormalTextRun SCXW54123816 BCX0\">f<\/span><\/span><span class=\"TextRun SCXW54123816 BCX0\" lang=\"EN-US\" xml:lang=\"EN-US\" data-contrast=\"auto\"><span class=\"NormalTextRun SCXW54123816 BCX0\">\u00a0you are\u00a0<\/span><\/span><span class=\"TextRun SCXW54123816 BCX0\" lang=\"EN-US\" xml:lang=\"EN-US\" data-contrast=\"auto\"><span class=\"NormalTextRun SCXW54123816 BCX0\">new to deep learning<\/span><\/span><span class=\"TextRun SCXW54123816 BCX0\" lang=\"EN-US\" xml:lang=\"EN-US\" data-contrast=\"auto\"><span class=\"NormalTextRun SCXW54123816 BCX0\">, feel free to leave the default value of padding.<\/span><\/span><span class=\"EOP SCXW54123816 BCX0\" data-ccp-props=\"{&quot;201341983&quot;:0,&quot;335559739&quot;:160,&quot;335559740&quot;:259}\">\u00a0<\/span><\/p>\n<p><span class=\"TextRun SCXW183928305 BCX0\" lang=\"EN-US\" xml:lang=\"EN-US\" data-contrast=\"auto\"><span class=\"NormalTextRun SCXW183928305 BCX0\">If\u00a0<\/span><\/span><span class=\"TextRun SCXW183928305 BCX0\" lang=\"EN-US\" xml:lang=\"EN-US\" data-contrast=\"auto\"><span class=\"NormalTextRun SCXW183928305 BCX0\">y<\/span><\/span><span class=\"TextRun SCXW183928305 BCX0\" lang=\"EN-US\" xml:lang=\"EN-US\" data-contrast=\"auto\"><span class=\"NormalTextRun SCXW183928305 BCX0\">ou\u2019d like to experiment with the tool to see the effect of changing the\u00a0<\/span><\/span><span class=\"TextRun SCXW183928305 BCX0\" lang=\"EN-US\" xml:lang=\"EN-US\" data-contrast=\"auto\"><span class=\"NormalTextRun SCXW183928305 BCX0\">p<\/span><\/span><span class=\"TextRun SCXW183928305 BCX0\" lang=\"EN-US\" xml:lang=\"EN-US\" data-contrast=\"auto\"><span class=\"NormalTextRun SCXW183928305 BCX0\">adding,\u00a0<\/span><\/span><span class=\"TextRun SCXW183928305 BCX0\" lang=\"EN-US\" xml:lang=\"EN-US\" data-contrast=\"auto\"><span class=\"NormalTextRun SCXW183928305 BCX0\">try running the tool on small areas with padding set to different\u00a0<\/span><\/span><span class=\"TextRun SCXW183928305 BCX0\" lang=\"EN-US\" xml:lang=\"EN-US\" data-contrast=\"auto\"><span class=\"NormalTextRun SCXW183928305 BCX0\">multiples of 8<\/span><\/span><span class=\"TextRun SCXW183928305 BCX0\" lang=\"EN-US\" xml:lang=\"EN-US\" data-contrast=\"auto\"><span class=\"NormalTextRun SCXW183928305 BCX0\">. In the graphic below, we are demonstrating how<\/span><\/span><span class=\"TextRun SCXW183928305 BCX0\" lang=\"EN-US\" xml:lang=\"EN-US\" data-contrast=\"auto\"><span class=\"NormalTextRun SCXW183928305 BCX0\"> a padding of<\/span><\/span><span class=\"TextRun SCXW183928305 BCX0\" lang=\"EN-US\" xml:lang=\"EN-US\" data-contrast=\"auto\"><span class=\"NormalTextRun SCXW183928305 BCX0\">64 px<\/span><\/span><span class=\"TextRun SCXW183928305 BCX0\" lang=\"EN-US\" xml:lang=\"EN-US\" data-contrast=\"auto\"><span class=\"NormalTextRun SCXW183928305 BCX0\"> is treated while inferencing.<\/span><\/span><span class=\"EOP SCXW183928305 BCX0\" data-ccp-props=\"{&quot;201341983&quot;:0,&quot;335559739&quot;:160,&quot;335559740&quot;:259}\">\u00a0<\/span><\/p>\n"},{"acf_fc_layout":"image","image":{"ID":1153532,"id":1153532,"title":"Padding in ArcGIS Pro","filename":"Padding-V7-1.gif","filesize":587922,"url":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2021\/03\/Padding-V7-1.gif","link":"https:\/\/www.esri.com\/arcgis-blog\/products\/arcgis-pro\/imagery\/deep-learning-with-arcgis-pro-tips-tricks-part-2\/padding-v7-2","alt":"Padding in ArcGIS Pro","author":"19291","description":"","caption":"","name":"padding-v7-2","status":"inherit","uploaded_to":1152172,"date":"2021-03-02 12:21:25","modified":"2021-03-02 12:21:38","menu_order":0,"mime_type":"image\/gif","type":"image","subtype":"gif","icon":"https:\/\/www.esri.com\/arcgis-blog\/wp-includes\/images\/media\/default.png","width":427,"height":240,"sizes":{"thumbnail":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2021\/03\/Padding-V7-1-213x200.gif","thumbnail-width":213,"thumbnail-height":200,"medium":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2021\/03\/Padding-V7-1.gif","medium-width":427,"medium-height":240,"medium_large":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2021\/03\/Padding-V7-1.gif","medium_large-width":427,"medium_large-height":240,"large":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2021\/03\/Padding-V7-1.gif","large-width":427,"large-height":240,"1536x1536":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2021\/03\/Padding-V7-1.gif","1536x1536-width":427,"1536x1536-height":240,"2048x2048":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2021\/03\/Padding-V7-1.gif","2048x2048-width":427,"2048x2048-height":240,"card_image":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2021\/03\/Padding-V7-1.gif","card_image-width":427,"card_image-height":240,"wide_image":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2021\/03\/Padding-V7-1.gif","wide_image-width":427,"wide_image-height":240}},"image_position":"center","orientation":"horizontal","hyperlink":""},{"acf_fc_layout":"content","content":"<p>For example, if the default is <strong>32 px<\/strong>, try running the tool with paddings of <strong>24 px<\/strong> and <strong>16<\/strong> <strong>px<\/strong> and compare the results. Check the images below to see the output of a model run with a padding of <strong>32 px<\/strong>\u00a0(in green) vs <strong>8 px<\/strong> (in purple).<\/p>\n"},{"acf_fc_layout":"image","image":{"ID":1152372,"id":1152372,"title":"Padding size: 32 Pixels","filename":"32Pad.png","filesize":584508,"url":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2021\/03\/32Pad.png","link":"https:\/\/www.esri.com\/arcgis-blog\/products\/arcgis-pro\/imagery\/deep-learning-with-arcgis-pro-tips-tricks-part-2\/32pad","alt":"Padding size: 32 Pixels","author":"19291","description":"","caption":"","name":"32pad","status":"inherit","uploaded_to":1152172,"date":"2021-03-01 15:34:14","modified":"2021-03-01 15:34:31","menu_order":0,"mime_type":"image\/png","type":"image","subtype":"png","icon":"https:\/\/www.esri.com\/arcgis-blog\/wp-includes\/images\/media\/default.png","width":986,"height":716,"sizes":{"thumbnail":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2021\/03\/32Pad-213x200.png","thumbnail-width":213,"thumbnail-height":200,"medium":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2021\/03\/32Pad.png","medium-width":359,"medium-height":261,"medium_large":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2021\/03\/32Pad.png","medium_large-width":768,"medium_large-height":558,"large":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2021\/03\/32Pad.png","large-width":986,"large-height":716,"1536x1536":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2021\/03\/32Pad.png","1536x1536-width":986,"1536x1536-height":716,"2048x2048":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2021\/03\/32Pad.png","2048x2048-width":986,"2048x2048-height":716,"card_image":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2021\/03\/32Pad-640x465.png","card_image-width":640,"card_image-height":465,"wide_image":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2021\/03\/32Pad.png","wide_image-width":986,"wide_image-height":716}},"image_position":"center","orientation":"horizontal","hyperlink":""},{"acf_fc_layout":"image","image":{"ID":1152382,"id":1152382,"title":"Padding size: 8 Pixels","filename":"8padding.png","filesize":597205,"url":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2021\/03\/8padding.png","link":"https:\/\/www.esri.com\/arcgis-blog\/products\/arcgis-pro\/imagery\/deep-learning-with-arcgis-pro-tips-tricks-part-2\/8padding","alt":"Padding size: 8 Pixels","author":"19291","description":"","caption":"","name":"8padding","status":"inherit","uploaded_to":1152172,"date":"2021-03-01 15:35:00","modified":"2021-03-01 15:35:29","menu_order":0,"mime_type":"image\/png","type":"image","subtype":"png","icon":"https:\/\/www.esri.com\/arcgis-blog\/wp-includes\/images\/media\/default.png","width":999,"height":724,"sizes":{"thumbnail":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2021\/03\/8padding-213x200.png","thumbnail-width":213,"thumbnail-height":200,"medium":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2021\/03\/8padding.png","medium-width":360,"medium-height":261,"medium_large":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2021\/03\/8padding.png","medium_large-width":768,"medium_large-height":557,"large":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2021\/03\/8padding.png","large-width":999,"large-height":724,"1536x1536":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2021\/03\/8padding.png","1536x1536-width":999,"1536x1536-height":724,"2048x2048":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2021\/03\/8padding.png","2048x2048-width":999,"2048x2048-height":724,"card_image":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2021\/03\/8padding-642x465.png","card_image-width":642,"card_image-height":465,"wide_image":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2021\/03\/8padding.png","wide_image-width":999,"wide_image-height":724}},"image_position":"center","orientation":"horizontal","hyperlink":""},{"acf_fc_layout":"content","content":"<h3>Batch size<\/h3>\n<p>Batch size is a term used in machine learning and refers to the number of image tiles the GPU can process at once while inferencing. The imagery is chopped up into tiles during inferencing, and the number of tiles the GPU can inference in one batch is called the batch size. If you run into out-of-memory errors with the tool, you need to reduce the batch size.<\/p>\n<p><span class=\"TextRun SCXW175924786 BCX0\" lang=\"EN-US\" xml:lang=\"EN-US\" data-contrast=\"auto\"><span class=\"NormalTextRun SCXW175924786 BCX0\">The batch size your computer can handle will depend on<\/span><\/span><span class=\"TextRun SCXW175924786 BCX0\" lang=\"EN-US\" xml:lang=\"EN-US\" data-contrast=\"auto\"><span class=\"NormalTextRun SCXW175924786 BCX0\">\u00a0the GPU available in your machine. To\u00a0<\/span><\/span><span class=\"TextRun SCXW175924786 BCX0\" lang=\"EN-US\" xml:lang=\"EN-US\" data-contrast=\"auto\"><span class=\"NormalTextRun SCXW175924786 BCX0\">determine<\/span><\/span><span class=\"TextRun SCXW175924786 BCX0\" lang=\"EN-US\" xml:lang=\"EN-US\" data-contrast=\"auto\"><span class=\"NormalTextRun SCXW175924786 BCX0\">\u00a0the\u00a0<\/span><\/span><span class=\"TextRun SCXW175924786 BCX0\" lang=\"EN-US\" xml:lang=\"EN-US\" data-contrast=\"auto\"><span class=\"NormalTextRun SCXW175924786 BCX0\">optimal<\/span><\/span><span class=\"TextRun SCXW175924786 BCX0\" lang=\"EN-US\" xml:lang=\"EN-US\" data-contrast=\"auto\"><span class=\"NormalTextRun SCXW175924786 BCX0\">\u00a0batch size, you m<\/span><\/span><span class=\"TextRun SCXW175924786 BCX0\" lang=\"EN-US\" xml:lang=\"EN-US\" data-contrast=\"auto\"><span class=\"NormalTextRun SCXW175924786 BCX0\">ay\u00a0<\/span><\/span><span class=\"TextRun SCXW175924786 BCX0\" lang=\"EN-US\" xml:lang=\"EN-US\" data-contrast=\"auto\"><span class=\"NormalTextRun SCXW175924786 BCX0\">need to run the tool a\u00a0<\/span><\/span><span class=\"TextRun SCXW175924786 BCX0\" lang=\"EN-US\" xml:lang=\"EN-US\" data-contrast=\"auto\"><span class=\"NormalTextRun SCXW175924786 BCX0\">few<\/span><\/span><span class=\"TextRun SCXW175924786 BCX0\" lang=\"EN-US\" xml:lang=\"EN-US\" data-contrast=\"auto\"><span class=\"NormalTextRun SCXW175924786 BCX0\">\u00a0times on a small geographical extent while\u00a0<\/span><\/span><span class=\"TextRun SCXW175924786 BCX0\" lang=\"EN-US\" xml:lang=\"EN-US\" data-contrast=\"auto\"><span class=\"NormalTextRun SCXW175924786 BCX0\">monitoring\u00a0<\/span><\/span><span class=\"TextRun SCXW175924786 BCX0\" lang=\"EN-US\" xml:lang=\"EN-US\" data-contrast=\"auto\"><span class=\"NormalTextRun SCXW175924786 BCX0\">your GPU metrics.\u00a0<\/span><\/span><span class=\"TextRun SCXW175924786 BCX0\" lang=\"EN-US\" xml:lang=\"EN-US\" data-contrast=\"auto\"><span class=\"NormalTextRun SCXW175924786 BCX0\">S<\/span><\/span><span class=\"TextRun SCXW175924786 BCX0\" lang=\"EN-US\" xml:lang=\"EN-US\" data-contrast=\"auto\"><span class=\"NormalTextRun SCXW175924786 BCX0\">tart your testing with a small batch size and increase the number as you go to maximize the GPU RAM usage.\u00a0<\/span><\/span><span class=\"TextRun SCXW175924786 BCX0\" lang=\"EN-US\" xml:lang=\"EN-US\" data-contrast=\"auto\"><span class=\"NormalTextRun SCXW175924786 BCX0\">We will cover how to monitor your GPU RAM usage in the \u201cRun the\u00a0<\/span><\/span><span class=\"TextRun SCXW175924786 BCX0\" lang=\"EN-US\" xml:lang=\"EN-US\" data-contrast=\"auto\"><span class=\"NormalTextRun SCXW175924786 BCX0\">Detect Objects Using Deep Learning<\/span><\/span><span class=\"TextRun SCXW175924786 BCX0\" lang=\"EN-US\" xml:lang=\"EN-US\" data-contrast=\"auto\"><span class=\"NormalTextRun SCXW175924786 BCX0\">\u00a0geoprocessing tool<\/span><\/span><span class=\"TextRun SCXW175924786 BCX0\" lang=\"EN-US\" xml:lang=\"EN-US\" data-contrast=\"auto\"><span class=\"NormalTextRun SCXW175924786 BCX0\">\u201d section below<\/span><\/span><span class=\"TextRun SCXW175924786 BCX0\" lang=\"EN-US\" xml:lang=\"EN-US\" data-contrast=\"auto\"><span class=\"NormalTextRun SCXW175924786 BCX0\">.<\/span><\/span><span class=\"EOP SCXW175924786 BCX0\" data-ccp-props=\"{&quot;201341983&quot;:0,&quot;335559739&quot;:160,&quot;335559740&quot;:259}\">\u00a0<\/span><\/p>\n<h3>Threshold<\/h3>\n<p>Prediction models output a level of confidence for each feature (i.e. building) they detect. The threshold parameter sets the minimum level of confidence that will be included in the output. For instance, if you set the threshold to <strong>0.9<\/strong>, any feature the model is less than 90% confident in will be eliminated. Another approach is to run the model with a lower threshold than your ultimate target, and then set a definition query on the output feature layer using the confidence attribute to display only features above a certain confidence threshold.<\/p>\n<h3>Return bounding box (return_bbox)<\/h3>\n<p>Return bounding box is a Boolean parameter with a <strong>True<\/strong> or <strong>False<\/strong> input. If <strong>True<\/strong>, the Detect Objects Using Deep Learning geoprocessing tool will return a bounding box around the detected feature rather than the feature itself.<\/p>\n<p>&nbsp;<\/p>\n<h2>Run Detect Objects Using Deep Learning<\/h2>\n<p>Now that you are familiar with the parameters for the Detect Objects Using Deep Learning tool, you are ready to start the inferencing process:<\/p>\n<ul>\n<li>From the <strong>Analysis<\/strong> tab on the main ribbon, click <strong>Tools<\/strong>.<\/li>\n<li>Under <strong>Find Tools<\/strong>, type <u>Detect Objects Using Deep Learning<\/u> and open the tool.<\/li>\n<li>Fill in the parameters as follows:\n<ul>\n<li><strong>Input Raster:<\/strong> Input your high resolution imagery.<\/li>\n<li><strong>Output Detected Object:<\/strong> Specify an output location for the detected features.<\/li>\n<li><strong>Model Definition:<\/strong> Import the previously downloaded dlpk.<\/li>\n<li><strong>padding:<\/strong> 32 (See details above or leave the default)<\/li>\n<li><strong>batch_size: <\/strong>16 (Leave the default if you have a Nvidia RTX 5000 GPU or equivalent. Decrease the batch size if you run into CUDA out-of-memory errors.)<\/li>\n<li><strong>threshold:<\/strong>9 (See details above or leave the default)<\/li>\n<li><strong>return_bboxes:<\/strong> False (See details above or leave the default)<\/li>\n<li>Check the<strong> Non Maximum Suppression <\/strong> This option is essential to merge overlapping detected features.<\/li>\n<li>Leave the other parameters as defaults. Do not run the tool yet.<\/li>\n<\/ul>\n<\/li>\n<\/ul>\n"},{"acf_fc_layout":"image","image":{"ID":1152402,"id":1152402,"title":"Detect Objects Using Deep Learning Geoprocessing Tool","filename":"DetectObjectsParameters.png","filesize":18145,"url":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2021\/03\/DetectObjectsParameters.png","link":"https:\/\/www.esri.com\/arcgis-blog\/products\/arcgis-pro\/imagery\/deep-learning-with-arcgis-pro-tips-tricks-part-2\/detectobjectsparameters","alt":"Detect Objects Using Deep Learning Geoprocessing Tool","author":"19291","description":"","caption":"","name":"detectobjectsparameters","status":"inherit","uploaded_to":1152172,"date":"2021-03-01 15:38:34","modified":"2021-03-01 15:39:03","menu_order":0,"mime_type":"image\/png","type":"image","subtype":"png","icon":"https:\/\/www.esri.com\/arcgis-blog\/wp-includes\/images\/media\/default.png","width":328,"height":583,"sizes":{"thumbnail":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2021\/03\/DetectObjectsParameters-213x200.png","thumbnail-width":213,"thumbnail-height":200,"medium":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2021\/03\/DetectObjectsParameters.png","medium-width":147,"medium-height":261,"medium_large":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2021\/03\/DetectObjectsParameters.png","medium_large-width":328,"medium_large-height":583,"large":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2021\/03\/DetectObjectsParameters.png","large-width":328,"large-height":583,"1536x1536":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2021\/03\/DetectObjectsParameters.png","1536x1536-width":328,"1536x1536-height":583,"2048x2048":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2021\/03\/DetectObjectsParameters.png","2048x2048-width":328,"2048x2048-height":583,"card_image":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2021\/03\/DetectObjectsParameters-262x465.png","card_image-width":262,"card_image-height":465,"wide_image":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2021\/03\/DetectObjectsParameters.png","wide_image-width":328,"wide_image-height":583}},"image_position":"center","orientation":"horizontal","hyperlink":""},{"acf_fc_layout":"content","content":"<ul>\n<li>Click the <strong>Environments\u00a0<\/strong>tab\n<ul>\n<li>If you are trying to determine the optimal batch size for your environment, zoom in to a small extent of your image where you can see about a dozen buildings and then set the <strong>Processing Extent <\/strong>to <strong>Current Display Extent<\/strong>. If you are confident in your batch size, leave the default Processing Extent settings.<\/li>\n<li>For <strong>Processor Type<\/strong>, pick <strong>GPU<\/strong>.<\/li>\n<li><strong>GPU ID: <\/strong>0 (for further explanation of this setting refer to the section in Part 1 of this blog series on installing CUDA).<\/li>\n<\/ul>\n<\/li>\n<li>Run the tool.<\/li>\n<\/ul>\n"},{"acf_fc_layout":"image","image":{"ID":1152412,"id":1152412,"title":"Detect Objects Using Deep LEarning Geoprocessing Tool Environment Variables","filename":"DetectObjectsEnvironments.png","filesize":20278,"url":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2021\/03\/DetectObjectsEnvironments.png","link":"https:\/\/www.esri.com\/arcgis-blog\/products\/arcgis-pro\/imagery\/deep-learning-with-arcgis-pro-tips-tricks-part-2\/detectobjectsenvironments","alt":"Detect Objects Using Deep LEarning Geoprocessing Tool Environment Variables","author":"19291","description":"","caption":"","name":"detectobjectsenvironments","status":"inherit","uploaded_to":1152172,"date":"2021-03-01 15:39:59","modified":"2021-03-01 15:40:23","menu_order":0,"mime_type":"image\/png","type":"image","subtype":"png","icon":"https:\/\/www.esri.com\/arcgis-blog\/wp-includes\/images\/media\/default.png","width":352,"height":623,"sizes":{"thumbnail":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2021\/03\/DetectObjectsEnvironments-213x200.png","thumbnail-width":213,"thumbnail-height":200,"medium":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2021\/03\/DetectObjectsEnvironments.png","medium-width":147,"medium-height":261,"medium_large":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2021\/03\/DetectObjectsEnvironments.png","medium_large-width":352,"medium_large-height":623,"large":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2021\/03\/DetectObjectsEnvironments.png","large-width":352,"large-height":623,"1536x1536":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2021\/03\/DetectObjectsEnvironments.png","1536x1536-width":352,"1536x1536-height":623,"2048x2048":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2021\/03\/DetectObjectsEnvironments.png","2048x2048-width":352,"2048x2048-height":623,"card_image":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2021\/03\/DetectObjectsEnvironments-263x465.png","card_image-width":263,"card_image-height":465,"wide_image":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2021\/03\/DetectObjectsEnvironments.png","wide_image-width":352,"wide_image-height":623}},"image_position":"center","orientation":"horizontal","hyperlink":""},{"acf_fc_layout":"content","content":"<p>To determine the optimal batch size for your GPU as discussed above, follow these steps as the tool is running:<\/p>\n<ul>\n<li>Open the command prompt window.<\/li>\n<li>Type <em><u>nvidia-smi<\/u><\/em> and press <em>Enter<\/em>.<\/li>\n<li>Under the <strong>Memory-Usage<\/strong> section, if you see there is some memory not being used, increase your batch size. If you see that the memory usage is at its max and the tool fails, decrease your batch size.<\/li>\n<\/ul>\n"},{"acf_fc_layout":"image","image":{"ID":1152422,"id":1152422,"title":"nvidia-smi Monitoring the GPU","filename":"MonitorGPU.png","filesize":38391,"url":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2021\/03\/MonitorGPU.png","link":"https:\/\/www.esri.com\/arcgis-blog\/products\/arcgis-pro\/imagery\/deep-learning-with-arcgis-pro-tips-tricks-part-2\/monitorgpu","alt":"nvidia-smi Monitoring the GPU","author":"19291","description":"","caption":"","name":"monitorgpu","status":"inherit","uploaded_to":1152172,"date":"2021-03-01 15:42:08","modified":"2021-03-01 15:42:29","menu_order":0,"mime_type":"image\/png","type":"image","subtype":"png","icon":"https:\/\/www.esri.com\/arcgis-blog\/wp-includes\/images\/media\/default.png","width":605,"height":313,"sizes":{"thumbnail":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2021\/03\/MonitorGPU-213x200.png","thumbnail-width":213,"thumbnail-height":200,"medium":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2021\/03\/MonitorGPU.png","medium-width":464,"medium-height":240,"medium_large":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2021\/03\/MonitorGPU.png","medium_large-width":605,"medium_large-height":313,"large":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2021\/03\/MonitorGPU.png","large-width":605,"large-height":313,"1536x1536":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2021\/03\/MonitorGPU.png","1536x1536-width":605,"1536x1536-height":313,"2048x2048":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2021\/03\/MonitorGPU.png","2048x2048-width":605,"2048x2048-height":313,"card_image":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2021\/03\/MonitorGPU.png","card_image-width":605,"card_image-height":313,"wide_image":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2021\/03\/MonitorGPU.png","wide_image-width":605,"wide_image-height":313}},"image_position":"center","orientation":"horizontal","hyperlink":""},{"acf_fc_layout":"content","content":"<p>Now that you know how to adjust the parameters to complete inferencing in ArcGIS Pro, you can iterate until you are happy with the output of the model. In the next blog post in this series, we will take this output and use ArcGIS Pro spatial analysis tools to further refine your result, including detecting and removing irregularities and misidentified features.<\/p>\n"}],"authors":[{"ID":19291,"user_firstname":"Rami","user_lastname":"Alouta","nickname":"Rami Alouta","user_nicename":"ralouta","display_name":"Rami Alouta","user_email":"RAlouta@esri.com","user_url":"","user_registered":"2020-03-23 16:57:54","user_description":"Rami is a Solution Engineer on the National Government team supporting nonprofit global organizations and land administration teams out of the Rotterdam office. He has over 5 years of GIS experience and has been working with Esri since 2016 previously as a Platform Configuration Engineer with Professional Services out of the Dubai office. He has a degree in Landscape Architecture from the American University of Beirut.","user_avatar":"<img data-del=\"avatar\" src='https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2020\/03\/0.jpg' class='avatar pp-user-avatar avatar-96 photo ' height='96' width='96'\/>"},{"ID":137201,"user_firstname":"Kate","user_lastname":"Hess","nickname":"Kate Hess","user_nicename":"khess","display_name":"Kate Hess","user_email":"khess@esri.com","user_url":"","user_registered":"2020-12-09 13:57:59","user_description":"Kate is a Business Development Manager on Esri's National Government team. Based in New York City, she has a background in remote sensing and environmental science. Kate currently supports National Statistics Offices globally, helping them modernize their census and statistics operations using GIS.","user_avatar":"<img data-del=\"avatar\" src='https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2026\/01\/IMG_8127-e1768502447568-213x200.jpeg' class='avatar pp-user-avatar avatar-96 photo ' height='96' width='96'\/>"}],"card_image":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2021\/03\/landclassification.jpg","wide_image":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2021\/03\/banner-2.jpg"},"yoast_head":"<!-- This site is optimized with the Yoast SEO Premium plugin v25.9 (Yoast SEO v25.9) - https:\/\/yoast.com\/wordpress\/plugins\/seo\/ -->\n<title>Deep Learning with ArcGIS Pro Tips &amp; Tricks: Part 2<\/title>\n<meta name=\"description\" content=\"Evaluate your imagery for deep learning and get started using out-of-the-box models such as building footprint extraction.\" \/>\n<meta name=\"robots\" content=\"index, follow, max-snippet:-1, max-image-preview:large, max-video-preview:-1\" \/>\n<link rel=\"canonical\" href=\"https:\/\/www.esri.com\/arcgis-blog\/products\/arcgis-pro\/imagery\/deep-learning-with-arcgis-pro-tips-tricks-part-2\" \/>\n<meta property=\"og:locale\" content=\"en_US\" \/>\n<meta property=\"og:type\" content=\"article\" \/>\n<meta property=\"og:title\" content=\"Deep Learning with ArcGIS Pro Tips &amp; Tricks: Part 2\" \/>\n<meta property=\"og:description\" content=\"Evaluate your imagery for deep learning and get started using out-of-the-box models such as building footprint extraction.\" \/>\n<meta property=\"og:url\" content=\"https:\/\/www.esri.com\/arcgis-blog\/products\/arcgis-pro\/imagery\/deep-learning-with-arcgis-pro-tips-tricks-part-2\" \/>\n<meta property=\"og:site_name\" content=\"ArcGIS Blog\" \/>\n<meta property=\"article:publisher\" content=\"https:\/\/www.facebook.com\/esrigis\/\" \/>\n<meta property=\"article:modified_time\" content=\"2021-10-07T16:27:18+00:00\" \/>\n<meta name=\"twitter:card\" content=\"summary_large_image\" \/>\n<meta name=\"twitter:site\" content=\"@ESRI\" \/>\n<script type=\"application\/ld+json\" class=\"yoast-schema-graph\">{\"@context\":\"https:\/\/schema.org\",\"@graph\":[{\"@type\":[\"Article\",\"BlogPosting\"],\"@id\":\"https:\/\/www.esri.com\/arcgis-blog\/products\/arcgis-pro\/imagery\/deep-learning-with-arcgis-pro-tips-tricks-part-2#article\",\"isPartOf\":{\"@id\":\"https:\/\/www.esri.com\/arcgis-blog\/products\/arcgis-pro\/imagery\/deep-learning-with-arcgis-pro-tips-tricks-part-2\"},\"author\":{\"name\":\"Rami Alouta\",\"@id\":\"https:\/\/www.esri.com\/arcgis-blog\/#\/schema\/person\/a8d8a386c26766f7e92a57c3eb7d3b20\"},\"headline\":\"Deep Learning with ArcGIS Pro Tips &amp; Tricks: Part 2\",\"datePublished\":\"2021-03-02T09:00:52+00:00\",\"dateModified\":\"2021-10-07T16:27:18+00:00\",\"mainEntityOfPage\":{\"@id\":\"https:\/\/www.esri.com\/arcgis-blog\/products\/arcgis-pro\/imagery\/deep-learning-with-arcgis-pro-tips-tricks-part-2\"},\"wordCount\":9,\"commentCount\":2,\"publisher\":{\"@id\":\"https:\/\/www.esri.com\/arcgis-blog\/#organization\"},\"keywords\":[\"ArcGIS Pro\",\"deep learning\",\"deep learning models\",\"geoAI\",\"Imagery\"],\"articleSection\":[\"Imagery &amp; Remote Sensing\",\"Mapping\"],\"inLanguage\":\"en-US\",\"potentialAction\":[{\"@type\":\"CommentAction\",\"name\":\"Comment\",\"target\":[\"https:\/\/www.esri.com\/arcgis-blog\/products\/arcgis-pro\/imagery\/deep-learning-with-arcgis-pro-tips-tricks-part-2#respond\"]}]},{\"@type\":\"WebPage\",\"@id\":\"https:\/\/www.esri.com\/arcgis-blog\/products\/arcgis-pro\/imagery\/deep-learning-with-arcgis-pro-tips-tricks-part-2\",\"url\":\"https:\/\/www.esri.com\/arcgis-blog\/products\/arcgis-pro\/imagery\/deep-learning-with-arcgis-pro-tips-tricks-part-2\",\"name\":\"Deep Learning with ArcGIS Pro Tips &amp; Tricks: Part 2\",\"isPartOf\":{\"@id\":\"https:\/\/www.esri.com\/arcgis-blog\/#website\"},\"datePublished\":\"2021-03-02T09:00:52+00:00\",\"dateModified\":\"2021-10-07T16:27:18+00:00\",\"description\":\"Evaluate your imagery for deep learning and get started using out-of-the-box models such as building footprint extraction.\",\"breadcrumb\":{\"@id\":\"https:\/\/www.esri.com\/arcgis-blog\/products\/arcgis-pro\/imagery\/deep-learning-with-arcgis-pro-tips-tricks-part-2#breadcrumb\"},\"inLanguage\":\"en-US\",\"potentialAction\":[{\"@type\":\"ReadAction\",\"target\":[\"https:\/\/www.esri.com\/arcgis-blog\/products\/arcgis-pro\/imagery\/deep-learning-with-arcgis-pro-tips-tricks-part-2\"]}]},{\"@type\":\"BreadcrumbList\",\"@id\":\"https:\/\/www.esri.com\/arcgis-blog\/products\/arcgis-pro\/imagery\/deep-learning-with-arcgis-pro-tips-tricks-part-2#breadcrumb\",\"itemListElement\":[{\"@type\":\"ListItem\",\"position\":1,\"name\":\"Home\",\"item\":\"https:\/\/www.esri.com\/arcgis-blog\/\"},{\"@type\":\"ListItem\",\"position\":2,\"name\":\"Deep Learning with ArcGIS Pro Tips &amp; Tricks: Part 2\"}]},{\"@type\":\"WebSite\",\"@id\":\"https:\/\/www.esri.com\/arcgis-blog\/#website\",\"url\":\"https:\/\/www.esri.com\/arcgis-blog\/\",\"name\":\"ArcGIS Blog\",\"description\":\"Get insider info from Esri product teams\",\"publisher\":{\"@id\":\"https:\/\/www.esri.com\/arcgis-blog\/#organization\"},\"potentialAction\":[{\"@type\":\"SearchAction\",\"target\":{\"@type\":\"EntryPoint\",\"urlTemplate\":\"https:\/\/www.esri.com\/arcgis-blog\/?s={search_term_string}\"},\"query-input\":{\"@type\":\"PropertyValueSpecification\",\"valueRequired\":true,\"valueName\":\"search_term_string\"}}],\"inLanguage\":\"en-US\"},{\"@type\":\"Organization\",\"@id\":\"https:\/\/www.esri.com\/arcgis-blog\/#organization\",\"name\":\"Esri\",\"url\":\"https:\/\/www.esri.com\/arcgis-blog\/\",\"logo\":{\"@type\":\"ImageObject\",\"inLanguage\":\"en-US\",\"@id\":\"https:\/\/www.esri.com\/arcgis-blog\/#\/schema\/logo\/image\/\",\"url\":\"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2018\/04\/Esri.png\",\"contentUrl\":\"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2018\/04\/Esri.png\",\"width\":400,\"height\":400,\"caption\":\"Esri\"},\"image\":{\"@id\":\"https:\/\/www.esri.com\/arcgis-blog\/#\/schema\/logo\/image\/\"},\"sameAs\":[\"https:\/\/www.facebook.com\/esrigis\/\",\"https:\/\/x.com\/ESRI\",\"https:\/\/www.linkedin.com\/company\/5311\/\"]},{\"@type\":\"Person\",\"@id\":\"https:\/\/www.esri.com\/arcgis-blog\/#\/schema\/person\/a8d8a386c26766f7e92a57c3eb7d3b20\",\"name\":\"Rami Alouta\",\"image\":{\"@type\":\"ImageObject\",\"inLanguage\":\"en-US\",\"@id\":\"https:\/\/www.esri.com\/arcgis-blog\/#\/schema\/person\/image\/\",\"url\":\"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2020\/03\/0.jpg\",\"contentUrl\":\"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2020\/03\/0.jpg\",\"caption\":\"Rami Alouta\"},\"description\":\"Rami is a Solution Engineer on the National Government team supporting nonprofit global organizations and land administration teams out of the Rotterdam office. He has over 5 years of GIS experience and has been working with Esri since 2016 previously as a Platform Configuration Engineer with Professional Services out of the Dubai office. He has a degree in Landscape Architecture from the American University of Beirut.\",\"sameAs\":[\"https:\/\/www.linkedin.com\/in\/rami-alouta-aa563790\/\"],\"url\":\"https:\/\/www.esri.com\/arcgis-blog\/author\/ralouta\"}]}<\/script>\n<!-- \/ Yoast SEO Premium plugin. -->","yoast_head_json":{"title":"Deep Learning with ArcGIS Pro Tips &amp; Tricks: Part 2","description":"Evaluate your imagery for deep learning and get started using out-of-the-box models such as building footprint extraction.","robots":{"index":"index","follow":"follow","max-snippet":"max-snippet:-1","max-image-preview":"max-image-preview:large","max-video-preview":"max-video-preview:-1"},"canonical":"https:\/\/www.esri.com\/arcgis-blog\/products\/arcgis-pro\/imagery\/deep-learning-with-arcgis-pro-tips-tricks-part-2","og_locale":"en_US","og_type":"article","og_title":"Deep Learning with ArcGIS Pro Tips &amp; Tricks: Part 2","og_description":"Evaluate your imagery for deep learning and get started using out-of-the-box models such as building footprint extraction.","og_url":"https:\/\/www.esri.com\/arcgis-blog\/products\/arcgis-pro\/imagery\/deep-learning-with-arcgis-pro-tips-tricks-part-2","og_site_name":"ArcGIS Blog","article_publisher":"https:\/\/www.facebook.com\/esrigis\/","article_modified_time":"2021-10-07T16:27:18+00:00","twitter_card":"summary_large_image","twitter_site":"@ESRI","schema":{"@context":"https:\/\/schema.org","@graph":[{"@type":["Article","BlogPosting"],"@id":"https:\/\/www.esri.com\/arcgis-blog\/products\/arcgis-pro\/imagery\/deep-learning-with-arcgis-pro-tips-tricks-part-2#article","isPartOf":{"@id":"https:\/\/www.esri.com\/arcgis-blog\/products\/arcgis-pro\/imagery\/deep-learning-with-arcgis-pro-tips-tricks-part-2"},"author":{"name":"Rami Alouta","@id":"https:\/\/www.esri.com\/arcgis-blog\/#\/schema\/person\/a8d8a386c26766f7e92a57c3eb7d3b20"},"headline":"Deep Learning with ArcGIS Pro Tips &amp; Tricks: Part 2","datePublished":"2021-03-02T09:00:52+00:00","dateModified":"2021-10-07T16:27:18+00:00","mainEntityOfPage":{"@id":"https:\/\/www.esri.com\/arcgis-blog\/products\/arcgis-pro\/imagery\/deep-learning-with-arcgis-pro-tips-tricks-part-2"},"wordCount":9,"commentCount":2,"publisher":{"@id":"https:\/\/www.esri.com\/arcgis-blog\/#organization"},"keywords":["ArcGIS Pro","deep learning","deep learning models","geoAI","Imagery"],"articleSection":["Imagery &amp; Remote Sensing","Mapping"],"inLanguage":"en-US","potentialAction":[{"@type":"CommentAction","name":"Comment","target":["https:\/\/www.esri.com\/arcgis-blog\/products\/arcgis-pro\/imagery\/deep-learning-with-arcgis-pro-tips-tricks-part-2#respond"]}]},{"@type":"WebPage","@id":"https:\/\/www.esri.com\/arcgis-blog\/products\/arcgis-pro\/imagery\/deep-learning-with-arcgis-pro-tips-tricks-part-2","url":"https:\/\/www.esri.com\/arcgis-blog\/products\/arcgis-pro\/imagery\/deep-learning-with-arcgis-pro-tips-tricks-part-2","name":"Deep Learning with ArcGIS Pro Tips &amp; Tricks: Part 2","isPartOf":{"@id":"https:\/\/www.esri.com\/arcgis-blog\/#website"},"datePublished":"2021-03-02T09:00:52+00:00","dateModified":"2021-10-07T16:27:18+00:00","description":"Evaluate your imagery for deep learning and get started using out-of-the-box models such as building footprint extraction.","breadcrumb":{"@id":"https:\/\/www.esri.com\/arcgis-blog\/products\/arcgis-pro\/imagery\/deep-learning-with-arcgis-pro-tips-tricks-part-2#breadcrumb"},"inLanguage":"en-US","potentialAction":[{"@type":"ReadAction","target":["https:\/\/www.esri.com\/arcgis-blog\/products\/arcgis-pro\/imagery\/deep-learning-with-arcgis-pro-tips-tricks-part-2"]}]},{"@type":"BreadcrumbList","@id":"https:\/\/www.esri.com\/arcgis-blog\/products\/arcgis-pro\/imagery\/deep-learning-with-arcgis-pro-tips-tricks-part-2#breadcrumb","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https:\/\/www.esri.com\/arcgis-blog\/"},{"@type":"ListItem","position":2,"name":"Deep Learning with ArcGIS Pro Tips &amp; Tricks: Part 2"}]},{"@type":"WebSite","@id":"https:\/\/www.esri.com\/arcgis-blog\/#website","url":"https:\/\/www.esri.com\/arcgis-blog\/","name":"ArcGIS Blog","description":"Get insider info from Esri product teams","publisher":{"@id":"https:\/\/www.esri.com\/arcgis-blog\/#organization"},"potentialAction":[{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https:\/\/www.esri.com\/arcgis-blog\/?s={search_term_string}"},"query-input":{"@type":"PropertyValueSpecification","valueRequired":true,"valueName":"search_term_string"}}],"inLanguage":"en-US"},{"@type":"Organization","@id":"https:\/\/www.esri.com\/arcgis-blog\/#organization","name":"Esri","url":"https:\/\/www.esri.com\/arcgis-blog\/","logo":{"@type":"ImageObject","inLanguage":"en-US","@id":"https:\/\/www.esri.com\/arcgis-blog\/#\/schema\/logo\/image\/","url":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2018\/04\/Esri.png","contentUrl":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2018\/04\/Esri.png","width":400,"height":400,"caption":"Esri"},"image":{"@id":"https:\/\/www.esri.com\/arcgis-blog\/#\/schema\/logo\/image\/"},"sameAs":["https:\/\/www.facebook.com\/esrigis\/","https:\/\/x.com\/ESRI","https:\/\/www.linkedin.com\/company\/5311\/"]},{"@type":"Person","@id":"https:\/\/www.esri.com\/arcgis-blog\/#\/schema\/person\/a8d8a386c26766f7e92a57c3eb7d3b20","name":"Rami Alouta","image":{"@type":"ImageObject","inLanguage":"en-US","@id":"https:\/\/www.esri.com\/arcgis-blog\/#\/schema\/person\/image\/","url":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2020\/03\/0.jpg","contentUrl":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2020\/03\/0.jpg","caption":"Rami Alouta"},"description":"Rami is a Solution Engineer on the National Government team supporting nonprofit global organizations and land administration teams out of the Rotterdam office. He has over 5 years of GIS experience and has been working with Esri since 2016 previously as a Platform Configuration Engineer with Professional Services out of the Dubai office. He has a degree in Landscape Architecture from the American University of Beirut.","sameAs":["https:\/\/www.linkedin.com\/in\/rami-alouta-aa563790\/"],"url":"https:\/\/www.esri.com\/arcgis-blog\/author\/ralouta"}]}},"text_date":"March 2, 2021","author_name":"Multiple Authors","author_page":"https:\/\/www.esri.com\/arcgis-blog\/products\/arcgis-pro\/imagery\/deep-learning-with-arcgis-pro-tips-tricks-part-2","custom_image":"https:\/\/www.esri.com\/arcgis-blog\/app\/uploads\/2021\/03\/banner-2.jpg","primary_product":"ArcGIS Pro","tag_data":[{"term_id":42181,"name":"ArcGIS Pro","slug":"arcgis-pro","term_group":0,"term_taxonomy_id":42181,"taxonomy":"post_tag","description":"","parent":0,"count":323,"filter":"raw"},{"term_id":186132,"name":"deep learning","slug":"deep-learning","term_group":0,"term_taxonomy_id":186132,"taxonomy":"post_tag","description":"","parent":0,"count":65,"filter":"raw"},{"term_id":757081,"name":"deep learning models","slug":"deep-learning-models","term_group":0,"term_taxonomy_id":757081,"taxonomy":"post_tag","description":"","parent":0,"count":3,"filter":"raw"},{"term_id":665211,"name":"geoAI","slug":"geoai","term_group":0,"term_taxonomy_id":665211,"taxonomy":"post_tag","description":"","parent":0,"count":36,"filter":"raw"},{"term_id":115262,"name":"Imagery","slug":"imagery","term_group":0,"term_taxonomy_id":115262,"taxonomy":"post_tag","description":"","parent":0,"count":152,"filter":"raw"}],"category_data":[{"term_id":22931,"name":"Imagery &amp; Remote Sensing","slug":"imagery","term_group":0,"term_taxonomy_id":22931,"taxonomy":"category","description":"","parent":0,"count":765,"filter":"raw"},{"term_id":22941,"name":"Mapping","slug":"mapping","term_group":0,"term_taxonomy_id":22941,"taxonomy":"category","description":"","parent":0,"count":2683,"filter":"raw"}],"product_data":[{"term_id":36581,"name":"ArcGIS Living Atlas","slug":"arcgis-living-atlas","term_group":0,"term_taxonomy_id":36581,"taxonomy":"product","description":"","parent":0,"count":1164,"filter":"raw"},{"term_id":36561,"name":"ArcGIS Pro","slug":"arcgis-pro","term_group":0,"term_taxonomy_id":36561,"taxonomy":"product","description":"","parent":0,"count":2035,"filter":"raw"}],"primary_product_link":"https:\/\/www.esri.com\/arcgis-blog\/?s=#&products=arcgis-pro","_links":{"self":[{"href":"https:\/\/www.esri.com\/arcgis-blog\/wp-json\/wp\/v2\/blog\/1152172","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.esri.com\/arcgis-blog\/wp-json\/wp\/v2\/blog"}],"about":[{"href":"https:\/\/www.esri.com\/arcgis-blog\/wp-json\/wp\/v2\/types\/blog"}],"author":[{"embeddable":true,"href":"https:\/\/www.esri.com\/arcgis-blog\/wp-json\/wp\/v2\/users\/19291"}],"replies":[{"embeddable":true,"href":"https:\/\/www.esri.com\/arcgis-blog\/wp-json\/wp\/v2\/comments?post=1152172"}],"version-history":[{"count":0,"href":"https:\/\/www.esri.com\/arcgis-blog\/wp-json\/wp\/v2\/blog\/1152172\/revisions"}],"wp:attachment":[{"href":"https:\/\/www.esri.com\/arcgis-blog\/wp-json\/wp\/v2\/media?parent=1152172"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.esri.com\/arcgis-blog\/wp-json\/wp\/v2\/categories?post=1152172"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.esri.com\/arcgis-blog\/wp-json\/wp\/v2\/tags?post=1152172"},{"taxonomy":"industry","embeddable":true,"href":"https:\/\/www.esri.com\/arcgis-blog\/wp-json\/wp\/v2\/industry?post=1152172"},{"taxonomy":"product","embeddable":true,"href":"https:\/\/www.esri.com\/arcgis-blog\/wp-json\/wp\/v2\/product?post=1152172"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}