Loading…
This layer is editable and shared with the public. To prevent unwanted editing, unshare this item or approve it for public data collection.
Finish setting up your layer
Describe your item below. Add fields on the Data tab. Configure editing on the Settings tab. Configure drawing and pop-ups through Map Viewer or Visualization tab.
Open
Creating Service

Deep learning model to classify point cloud data into building or background classes. A brief summary of the item is not available. Add a brief summary about the item.

‎Deep learning package by

New notebook runtime available. You can update the runtime from the settings tab of the item details page.

Item created: Jun 24, 2024 Item updated: Dec 31, 2024 Number of downloads: 1,158

Snapshot last refreshed:

1963 characters left.

Description

The classification of point cloud datasets to identify points that belong to a building class is a fundamental GIS use case. Classification of building points is a key step in creating 3D models/digital twins and in change detection workflows for infrastructure projects. The classification of building points is also crucial in applications where buildings' vicinity from objects of interest such as power poles, wires, trees, and so on is needed to be determined.  

The model was trained on airborne lidar datasets, without high or low noises and is expected to perform best with similar datasets. Classification of terrestrial point cloud datasets may work but has not been validated. For these or similar cases, this pretrained model may be fine-tuned to save on cost, time, and compute resources while improving accuracy. When fine-tuning this model, the target training data characteristics such as class structure, maximum number of points per block, and extra attributes should match those of the data originally used for training this model (see Training data section below).

Using the model
Follow the guide to use the model. The model can be used with ArcGIS Pro's Classify Point Cloud Using Trained Model tool. Before using this model, ensure that the supported deep learning libraries are installed. For more details, check Deep Learning Libraries Installer for ArcGIS.

Input

The model accepts point clouds with point geometry (X, Y and Z values). 


Note: The model is not dependent on any additional attributes such as Intensity, Number of Returns, etc. The model was trained using a training dataset with the full set of points. Therefore, it is important to make the full set of points available to the neural network while predicting - allowing it to better discriminate points of 'class of interest' versus background points. It is recommended to use 'selective/target classification' and 'class preservation' functionalities during prediction to have better control over the classification and scenarios with false positives.


Output

The model will classify the point cloud into the following classes as per their meaning defined by the American Society for Photogrammetry and Remote Sensing (ASPRS) :

0
Background
6Building



Applicable geographies
The model is expected to work within any geography. However, results can vary for datasets that are statistically dissimilar to training data.

Model architecture
This model uses the RandLANet model architecture implemented in ArcGIS API for Python.


Training data

This model is trained on a subset of Canada & Netherland's open dataset. The training data used has the following characteristics:

 X, Y, and Z linear unitmeter
 Z range-168.48 m to 232.96 m
 Number of Returns
1 to 7
 Point spacing0.1 to 0.6
 Block size100 m
 Maximum points per block30000
 Extra attributesNone
 Class structure[0, 6]

Limitations
Predictions for skyscrapers and very large warehouses can be less precise as compared to standalone or connected small-to-medium sized buildings.

Sample results
Here are a few results from the model.

results 1


result 2

An in-depth description of the item is not available.

Layers

Ground Layers

Tools

Tables

Basemap

Project Contents:

Solution Contents

Contents

Layers

Screenshots

Terms of Use

https://downloads.esri.com/blogs/arcgisonline/esrilogo_new.png
This work is licensed under the Esri Master License Agreement.

No special restrictions or limitations on using the item's content have been provided.

Comments (2)

Sign in to add a comment.

siegel_j Item Owner commented 7 months ago Delete Reply

Link to the guide is broken

vraj_deldev Item Owner commented 7 months ago Delete

Fixed, thanks for reporting it.

Item Information

LowHigh

Item Information

LowHigh

Make your item easy to find, understand, and use by providing this information.

    Details

    Dashboard views: Desktop

    Source: (Image Collection)

    Creating data in:

    Published as:

      Other Views:

        Dependent items in the recycle bin

          Applicable: 2d

          Size: 5.832 MB

          Attachments size: 0 KB

          ID: a64fa0b01aef406c8a0b2feaa1feee76

          Image Count: 0

          Image Properties

          Layer Drawing

          Using tiles from a cache

          Dynamically from data

          Share
          Owner

          esri_analytics

          Folder

          Categories

          This item has not been categorized.

          Assign Category
          Tags

          LivingAtlasDLPK, dlpk, Point Cloud Classification, point cloud, deep learning, living atlas

          Edit Tags
          Credits (Attribution)
          No acknowledgements.

          Esri

          ArcGIS Navigator link
          URL View
          WMTS View
          Your tile layer is ready to use
          This tile layer will automatically create tiles as needed and cache them for future use. No further configuration is required. View the Settings tab to change the default options. Build tiles manually for specific scales and extents to improve display performance for the first person to view the tile layer at that scale and extent. Tiles must exist if the layer will be used offline.
          All items were exported successfully
          ${numberOfItems} item(s) were exported successfully. Some item(s) skipped or failed to export.
          See description for more information
          Cannot import
          Export packages from newer portal versions cannot be imported to older versions.
          Loading…