What's new in version 1.8.0

New

  • Uses the requests 3rd party library instead of the built-in urllib for all web requests. Benefits include:
    • increased security
    • standardized web requests
    • integrated libraries for better PKI, IWA, and other authentication
  • Adds ability to install ArcGIS API for Python into Python 3.8 conda environments
  • Certifies use of Pandas releases greater than 1.x

Guides

Samples

Module Changes

arcgis.gis

arcgis.gis.admin

  • Adds missing parameters to the ArcGIS Online history function
  • all_events, event_ids, event_types, actors, owners, actions, ips, sort_order, data_format

arcgis.gis.nb

arcgis.features

arcgis.raster

arcgis.raster.functions

  • Adds code sample to con function documentation

arcgis.raster.analytics

arcgis.geoanalytics

  • Adds new parameters to the find_point_clusters tool
  • time_method
  • search_duration
  • duration_unit

arcgis.geoprocessing

  • Adds choice_list attribute on the module returned from import_toolbox() to view the options list for each parameter

arcgis.mapping

arcgis.widgets

arcgis.learn

Fixes

Guides

Samples

Module Changes

arcgis.gis

  • Fixes clone_items() between Enterprise 10.4.1 and 10.5.1
  • Updates clone_items() to handle thumbnails in all situations
  • Fixes BUG-000128343 Group.update() method not updating when is_invitation_only set to False
  • Improves error handling for various get() functions when item does not exist or is inaccessible

arcgis.gis.admin

arcgis.gis.nb

arcgis.features

arcgis.features.analysis

arcgis.mapping

arcgis.learn

  • Enhances module to handle training set images with missing xml files
  • Fixes save() method framework AttributeError for arcgis.learn models
  • Improves reporting of accuracy metrics for models
  • Fixes issue where validation loss curve was not displayed when traning for more than one epoch
  • Fixes AttributeError on user-defined learning rate
  • Fixes DeepLab model when using multispectral imagery and the ResNet101 backbone
  • Fixes Retinanet.predict_video() function when resize parameter is not set
  • Fixes MaskRCNN.from_model() function to properly set default backbone when inferencing
  • Fixes DeepLab.fit() and PSPNet.fit() methods run on objects created with respective from_model() methods
  • Fixes prepare_data() when resize_to parameter is larger than chip_size causing reflection padding and scale the image to an incorrect size
  • Fixes prepare_data() chip_size on objects prepared with transforms=False and resize_to=<int>

Your browser is no longer supported. Please upgrade your browser for the best experience. See our browser deprecation post for more details.