What's new in version 1.8.3

New

  • Adds ujson library as a dependency for JSON parsing to improve performance

New Guides

New Samples

arcgis.gis

arcgis.gis.admin

arcgis.gis.server

arcgis.env

  • Adds environment variables:
  • union_dimension
  • match_variables

arcgis.features

arcgis.features.analysis

arcgis.features.managers

arcgis.raster

arcgis.raster.analytics

  • Enhances API to utilize the hosted imagery and raster analysis capabilities available with the June 2020 ArcGIS Online
  • Adds tiles_only keyword argument to functions
  • build_multidimensional_transpose()
  • Adds new parameter:
  • delete_transpose (Parameter available in ArcGIS Image Server 10.9 or higher)
  • summarize_raster_within()
  • Adds new parameter:
  • percentile_interpolation_type(Parameter available in ArcGIS Image Server 10.9 or higher)
  • Enhanced the following functions to accept local datasets to create hosted imagery layers
  • create_image_collection()
  • copy_raster()
  • Adds merge_multidimensional_rasters() function (requires ArcGIS Image Server 10.9 or higher)
  • calculate_density()
  • Adds input_barriers parameter (requires in ArcGIS Image Server 10.9 and higher)
  • optimal_path_as_line()
  • Adds create_network_paths parameter (requires in ArcGIS Image Server 10.9 and higher)

arcgis.raster.functions

  • Adds new functions:
  • detect_change_using_change_analysis_raster
  • trend_to_rgb
  • apparent_reflectance
  • buffered
  • rasterize_features
  • reproject
  • heat_index
  • wind_chill
  • aspect_slope
  • contour
  • compute_change
  • aggregate
  • bai
  • nbr
  • ndbi
  • ndmi
  • ndsi
  • mndwi
  • band_arithmetic()
  • Adds new options for method:
    • BAI
    • NBR
    • NDBI
    • NDMI
    • NDSI
    • MNDWI
  • generate_trend()
  • Adds options for regression_type parameter:
    • MANN-KENDALL
    • SEASONAL-KENDALL
  • Adds new parameter:
    • seasonal_period
  • focal_statistics()
  • Adds new options for stat_type:
    • Median
    • Majority
    • Minority
  • composite_band()
  • Adds cellsize_type parameter
  • geometric()
  • Adds new parameters:
    • tolerance
    • dem

arcgis.raster.functions.gbl

  • Adds new functions:
  • boundary_clean() (Parameter available in ArcGIS Image Server 10.9 or higher)
  • zonal_statistics()
  • Adds new parameter:
  • percentile_interpolation_type (Parameter available in ArcGIS Image Server 10.9 or higher)
  • kernel_density()
  • Adds in_barriers parameter (Parameter available in ArcGIS Image Server 10.9 and higher)

argis.raster.functions.RFT

  • Enhanced module to work with ArcGIS Online

arcgis.geoanalytics

  • summarize_data submodule
    • Adds time stepping parameters to summarize_attributes() for 10.9+ releases
      • time_step_interval
      • time_step_interval_unit
      • time_step_repeat_interval
      • time_step_repeat_interval_unit
      • time_step_reference

arcgis.geometry.functions

  • Adds missing api reference documentation

arcgis.mapping

arcgis.mapping.ogc

  • GeoJSONLayer
  • Adds data parameter to allow plotting from a string or local file

arcgis.widgets

arcgis.apps

arcgis.learn

  • Adds ability to save model_metric.html
  • Adds time spent per epoch
  • Adds extension to support transformer models
  • Adds f1_score() description to API reference
  • adds default transformations
  • FasterRCNN
  • Adds multispectrals support when tensorboard=True
  • Adds predict() method
  • Adds predict_video()
  • Adds support for kwargs parameter
  • MLModel
  • Adds support for unsupervised learning
  • Clustering Models
  • Gaussian Mixture Models
  • Novelty and Outlier Detection Models
  • Adds methods:
  • decision_function()
  • feature_importances()
  • predict()
  • renames output_raster_folder_path to output_raster_path
  • renames predict_features to prediction_type
  • Adds multiband raster support
  • FeatureClassifier
  • Adds multi-label support for training
  • Adds note to oversample parameter explaining supported datasets
  • Adds parameters to save():
    • compute_metrics
    • save_optimizer
    • save_inference_file
  • Adds note to predict() that it only supports RGB images
  • Adds PointRend to improve performance for:
  • DeepLab
  • PSPNetClassifer
  • MaskRCNN
  • PointCNN
  • Updates how transforms are applied on point cloud so both original data and trasformed data passed
  • Updates dependencies in SciPy and removes dependencies on transform3d and plotly-orca
  • Adds ability to preserve classcodes in file when using predict_las()

Fixes

Guides

Samples

arcgis.gis

  • Fixes parameter names in PortalDataStore.publish() code example for the config parameter
  • Fixes authentication error with Python 3.7 and Notebook Server
  • Clarifies documentation for storageUsage parameter of User objects
  • Fixes issue with ArcGIS online comments with certain unicode characters not getting returned properly
  • Fixes error with clone_items() on Survey items
  • Fixes deprecation warnings when import API in Notebook Server
  • Item
  • Fixes documentation error in parameter table for reassign_to()
  • UserManager
  • Fixes KeyError: UserTypes error with create() when user_settings configured improperly
  • Fixes token errors with public Geometry and Geoprocessing services affecting API and Notebook Server
  • Fixes BUG-000133849 - data resource is missing when KML or KMZ files are added to ArcGIS Online
  • ContentManager
  • Fixes BUG-000133739 - clone_items() on hosted feature layer views returns JSONDecodeError: Expecting value: line 1 column 1 (char 0) error
  • GroupMigrationManger
  • Fixes to inspect() and load() not recoginizing new items

arcgis.gis.admin

  • Fixes error returning ArcGIS Online history() when return type is a DataFrame

arcgis.gis.server

  • Fixes issue where a "/" added to the server admin url creates invalid connection
  • Fixes issue when creating Server object without a Portal connection on a Federated Server
  • Fixes issue with mangled URL when connecting to unfederated servers URL
  • ReportManager
  • BUG-000133053 list() method returns error if a saved report created from ArcGIS Server manager exists on server
  • Fixes issue when create() fails if existing reports named as integers
  • Fixes errors in create() documentation for parameters:
    • queries
    • from_value
    • to_value
  • Fixes BUG-000134098 unexpected keyword argument from_value error in quick_report()
  • Removes incorrect option from since parameter documentation for quick_report()

arcgis.features

arcgis.features.analysis

arcgis.features.managers

arcgis.raster

arcgis.raster.functions

arcgis.network

  • Sets default values for optional parameters to None to rely on service defaults whether in ArcGIS Online and ArcGIS Enterprise

arcgis.geoanalytics

arcgis.geocoding

  • Updates suggest() so location parameteris optional

arcgis.geometry

  • Fixes issue with various functions ignoring inplace paramter default value
  • Fixes issue with various functions not returning the spatialReference with output
  • Fixes issue with Polygon failing if order of geometries used is not correct for EsriJSON format

arcgis.mapping

  • Fixes BUG-000131799 - import_tiles() returns error:"'Response' object is not subscriptable" because of improper url construction
  • Fixes error when extent parameter for OfflineMapAreaManager.create() is a bookmark or dictionary
  • WebMap
  • Fixes AttributeError: layerType error when calling save()
  • Fixes error on save() if WebMap object had a FeatureLayerColletion added with add_layer()
  • Fixes issue with add_layer() where only layers are added from an an Item or FeatureLayerCollection that has both layers and tables
  • Fixes issue with add_table() trying to get a renderer when adding
  • Fixes BUG-000131725 - complex symbology is not preserved for Map Service added using the add_layer() function
  • Fixes issue with zoom_to_layer() function failing to display layer if called before map has fully loaded

arcgis.widgets

arcgis.apps

arcgis.learn

  • Disables saving optimizer by default (default value of save_optimizer parameter False in models
  • Numerous documentation enhancements
  • Fixes fit() methods on multiple models when using tensorboard=True
  • Fixes bug with prepare_data() failing with Multispectral data when gdal is not installed
  • Removes unnecessary warning messages when debugging is set
  • Fixes issue where classcodes are not preserved in output raster when training data has non-contiguous classcode labels for Pixel Classification models
  • Fixes issue with inferencing when prepare_data() dataset_type parameter is ChangeDetection and training data is multispectral
  • detect_objects()
  • Fixes warning messages when deciding between GPU or CPU
  • SuperResolution
  • Fixes prepare_data issue when not creating labels when called without downsampling_factor
  • fixes issue with default transforms causing prediction errors
  • fixes error when image data format is tiff
  • fixes chip_size causing artifacts while inferencing
  • Fixes errors when gdal was not handling non-contiguous class segementation
  • EntityRecognizer
  • Fixes issues with large loss values during model training
  • Fixes per epoch metric calculation bug
  • Fixes lr_find() to run for at least one full epoch per learning rate
  • Fixes issue with lr_find() output when data prepared with batch_size of 2
  • Fixes text column to be first column in show_batch()
  • FeatureClassifier
  • Fixes issue with fit() not displaying results after training with multipsectral data
  • Fixes issues with show_results() failing with smaller validation datasets
  • Fixes failures of plot_hard_examples() with multilabel model
  • Fixes errors when running fit() if oversample=True
  • Fixes issues with plot_losses()
  • UnetClassifier
  • Fixes issue with incorrect exceptions raised when object craeted with no data
  • Fixes missing model type information from output of save()
  • Fixes ValueError with lr_find() and fit()
  • Fixes show_results() tesnor shape error
  • Fixes class weights not used even after specifying class_balancing=True
  • Fixes class weights not matching number of classes when data has been exported with one class having pixel value 0
  • Fixes inconsistent accuracy() results between fitting and loading
  • PSPNetClassifier
  • Fixes missing model type information from output of save()
  • Fixes inconsistent accuracy() results between fitting and loading
  • Fixes ValueError with lr_find() and fit()
  • Fixes NameError when intializing with use_unet and unet_aux_loss both set to True
  • Fixes class weights not used even after specifying class_balancing=True
  • Fixes class weights not matching number of classes when data has been exported with one class having pixel value 0
  • Fixes error with lr_find() and VGG backbones if model created with use_unet=True
  • Fixes errors in fit() on object created wtih VGG backbone and pointrend=True
  • DeepLab
  • Fixes class weights not used even after specifying class_balancing=True
  • Fixes class weights not matching number of classes when data has been exported with one class having pixel value 0
  • Fixes inconsistent accuracy() results between fitting and loading
  • MaskRCNN
  • Fixes situation with incorrect labels when multiband data is exported
  • FasterRCNN
  • Fixes show_results() failing with multispectral data
  • Fixes issue with save() raising exception instead of returning no boxes when using default parameters
  • Fixes fit() method when tensorboard=True
  • RetinaNet
  • Fixes to auto-detection routines to determine CPU or GPU hardware available
  • PointCNN
  • Fixes predict_las() failing when output folder is not empty or has files with pred in name or .las extension
  • Fixes issues with las and h5 files remaining open after using
  • Fixes issue predict_las() failing if .las files have a period in the filename
  • Multiple fixes to documentation
  • SingleShotDetector
  • Fixes predict() TypeError when using multispectral data
  • Fixes errors when initializing with mobilnet_v2 or densenet backbones
  • Fixes errows with predict() when using multipsectral data and visualize=True
  • EntityRecognizer
  • Fixes issues with incorrect exceptions raised caling fit() or lr_find() if object initialized without data
  • Fixes issue where saving checkpoint was not using the most apt value of validation loss when early_stopping is set to true
  • Fixes issue where the HTML file with model metrices (model_metrices.html) not getting saved when training UnetClassifer, PSPNetClassifier, DeepLab, SingelShotDetector and RetinaNet models

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