What's new in version 1.9.0

• Adds support for ArcGIS Enterprise on Kubernetes
• Adds standardized output_names for all arcgis.geoanalytics and arcgis.features.analysis analysis tools
• Updates the map widget to use JavaScript 4.19
• Adds classes to manage ArcGIS Velocity functionality
• Enhances API Reference for the ArcGIS API for Python
  • Adds text emphasis, links to class documentation when objects are returned, and code snippets:
    • arcgis.features
    • arcgis.geometry
    • arcgis.widgets

New

Guides

• Imagery and raster analysis
  • New Guides:
    • Create Imagery Layers
• Deep Learning with ArcGIS
  • Updates guides for default unfrozen backbone in text models
  • New Guides:
    • Track objects using SiamMask
    • Multi-object Tracking using ObjectTracker
  • [Inference only Text Models in arcgis.learn](https://developers.arcgis.com/python/guide/inference-only-text-models-in-arcgis-learn/)
    • Adds Choosing a different backbone than the default section and various text edits
• Geocoding guides
  • Part 1 - What is geocoding?
  • Part 2 - Locating addresses
  • Part 3 - Finding points of interest
  • Part 4 - Batch geocode
  • Part 5 - Reverse geocode
  • Part 6 - Custom geocoders
  • Part 7 - Utility functions for geocoding

Samples

• Updates Runtime used in Try It Live Notebooks to 5.0
• GIS analysts and data scientists
  • Updates samples for default unfrozen backbone in text models
  • New Samples:
    • Determining site suitability for oil palm plantation
    • Model explainibility for ML Models
    • Supervised learning of tabular data using AutoML
  • Updates symbols used in Identifying suitable sites for new ALS clinics using location allocation analysis
  • Updates SAR to RGB image translation using CycleGAN to include additional explanation and minor modifications to functions
• Content publishers
  • Updates to Publishing web maps and web scenes

Module Changes

arcgis.gis

• GIS
  • Adds notebook_server property to return the NotebookServer registered with the organization
  • Adds Usage Examples for connecting with token and api_key parameters
  • Adds Note to map() documentation that IE 11 is no longer supported
  • Adds note on certificate path to verify_cert parameter
  • Adds velocity property to manage ArcGIS Velocity functionality
  • Adds properties:
    • servers
    • languages
    • regions
• Item
  • Enhances usage() method to allow entering custom date ranges
  • Adds ability to publish OGC Feature Service from existing Feature Layer item using the output_type parameter
  • Adds snapshots property to return notebook item snapshots
  • Adds can_delete property to check for ability to remove item
• InvitationManager
  • Adds manage_invitations method to allow users to accept or deny invitations by invitationId
• ContentManager
  • Adds preserve_item_id parameter to clone_items() to keep the same item_id when target organization is ArcGIS Enterprise 10.9 or later
• User
  • Adds culture_format parameter to update() method

arcgis.gis.admin

• AGOLAdminManager
  • Adds scheduled_tasks() method to allow org_admins to view scheduled tasks

arcgis.gis.server

• Adds documentation for:
  • JobManager
  • Job

arcgis.gis.mission

• Adds documentation for:
  • Mission
  • MissionCatalog
  • MissionJob

arcgis.gis.nb

• Adds classes to module:
  • SnapshotManager class to manage Notebook Snapshots
  • SnapShot class to represent individual notebook item snapshot
  • SystemManager
  • DirectoryManager
  • Container
  • ContainerNotebook
  • WebAdaptor
  • WebAdaptorManager
  • LogManager
  • SiteManager
  • MachineManager
  • Machine
  • SecurityManager

arcgis.env

• Adds WKT support to process_spatial_reference and out_spatial_reference

arcgis.features

• Adds internal job status checks for asynchronous jobs that impact many methods in the module
• FeatureLayer
  • Adds incremental checking of job status to append() when run asynchronously
  • Adds required to the source_table_name parameter in the API Reference documentation for append()
  • Adds capability to run following methods asynchronously:
    • append()
  • Updates how edit_features() sends request with large amounts of data
  • Adds support for a list of feature objects as adds argument to edit_features()
• Spatially Enabled DataFrame
  • Adds ValueError to from_featureclass() if location argument is not correct format

arcgis.features.managers

• FeatureLayerCollectionManager
  • Adds parameters to create_view()
    • overwrite
    • description
    • tags
    • snippet
    • set_item_id
  • Adds ability to run following functions asynchronously:
    • add_to_definition()
    • delete_from_definition()
    • update_definition()
  • Improves error messaging with overwrite() if outstanding replicas exist
• FeatureLayerManager
  • Adds ability to run following functions asynchronously:
    • add_to_definition()
    • delete_from_definition()
    • update_definition()
• ParcelFabricManager
  • Adds divide() method
• AttachmentManager
  • Adds parameters to the search() method
    • return_metadata
    • return_url
    • max_records
    • offset

arcgis.raster

• Adds numerous code snippets, documentation improvements and corrections to module and all submodules
• Adds support for publishing a single image
• Adds support for SpatioTemporal Asset Catalogs
  • Raster
    • create Raster objects from stac items with from_stac_item()
  • RasterCollection
    • create RasterCollection objects from stac api search queries with from_stac_api()
    • create RasterCollection objects from stac catalogs with from_stac_catalog()
• ImageryLayer
  • Adds plot_histograms() method
  • Updates nodata values to appear white with tiles only services
  • Updates export_image() to include numpy_array as an option for the f argument
  • Adds full support for creating Dynamic Imagery Layers with the save(tiles_only=False) argument

arcgis.raster.analytics

• Adds full support for creating Dynamic Imagery Layers to ArcGIS Online with tiles_only=False argument with those functions accepting it
• aggregate_multidimensional_raster()
  • Corrects example for interval_ranges parameter
  • New parameters:
    • dimensionless
    • percentile_value
    • percentile_interpolation_type
  • Adds PERCENTILE as option for aggregation_method argument
• Adds following to create_image_collection() and copy_raster()
  • Updates context parameter documentation with upload_properties key for parallel processing and progress bar
  • Adds Tiled Imagery Layer as an option for the raster_type_name argument
  • Adds numerous Usage Examples to the reference documentation
• create_image_collection()
  • Removes tiles_only parameter

arcgis.raster.functions

• Adds to aggregate():
  • PERCENTILE as option for aggregate_function argument
  • New parameters:
    • dimensionless
    • percentile_value
    • percentile_interpolation_type
• Adds to max()
  • ignore_nodata parameter

arcgis.raster.functions.gbl

• Adds function:
  • viewshed()
• Deprecates least_cost_path() function

arcgis.raster.utils

• upload_imagery_to_agol_userstore()
  • Adds documentation and code snippet
  • Adds parameters:
    • auto_renew
    • upload_properties - to support parallel processing and provide option for progress bar

arcgis.network

• Updates how methods handle setting default travel_mode when not supplied

arcgis.geoanalytics

• Adds WKT support for context parameter in all functions

• summarize_data

  • Adds keep_target parameter to join_features()
  • Adds parameters to reconstruct_tracks()
    • arcade_split
    • split_boundary

• manage_data

  • overlay_data()
    • Updates include_overlaps parameter documentation
  • append_data()
    • Adds code sample for the field_mappings parameter

arcgis.mapping.ogc

• Adds submodule documentation:
  • arcgis.mapping.forms
• Adds new classes:
  • OGCFeatureService to represent an OGC Feature Server
  • OGCCollection to represent an individual OGC dataset
• WebMap
  • Adds reference documentation for:
    • forms property
    • print() method

arcgis.realtime

• Adds new classes:
  • Velocity
  • FeedsManager
  • RealTimeAnalyticsManager
  • BigDataAnalyticsManager
  • Feed
  • RealTimeAnalytics
  • BigDataAnalytics

arcgis.learn

• Enhances training (fit() method) for all models to add the average precision score as an option for the monitor parameter

• Enhances available_metrics property to report in ArcGIS Pro

• Adds support for loading pre-trained Imagery models in an offline mode

• Adds Experimental support for tflite as the framework for:

  • FeatureClassifier
  • SingleShotDetector
  • RetinaNet

• Data Preparation Methods

  • Adds capability to select specific raster bands in prepare_tabulardata() by using the explanatory_rasters parameter

• Object Classification Models

  • Adds support for custom DataBunch

• Object Detection Models

  • New Models
    • MMDetection

• Feature, Tabular, and TimeSeries Models

  • Adds note to predict() and predict_video() methods that support is only for RGB images
  • New Models
    • AutoML
  • Adds SHAP Interpretability to predict() with new explain and explain_index parameters
    • FullyConnectedNetwork
    • MLModel
  • MLModel
    • Adds publishing predict(prediction_type="raster") output directly to organization rather than locally

• Change Inferencing Methods (Image Server) to Inferencing Methods

  • New Class
    • Embeddings

• Pixel Classification Models

  • New Models
    • MMSegmentation

• Unstructured Text Models

  • arcgis.learn.text module
    • Adds rows and max_len parameters to the show_batch() utility function
    • Adds freeze() method to transformer based models
    • EntityRecognizer
      • adds support for custom labels
      • adds allow_plot documentation for lr_find()
      • adds show_progress parameter to extract_entities() method
      • removes using FastTokenizers from transformer backbones
    • Inferencing Only Models
      • adds pretrained_path and working_dir parameters as keyword argument options

• Model Management

  • Adds warning to train_model() that it is not supported for ArcGIS Online
  • Updates train_model() to process asynchronously

• Image Translation Models

  • Adds support for loading models that were saved with save_inference_file argument set to False
  • ImageCaptioner
    • Adds additional documentation to framework parameter of the save() method
  • Pix2Pix
    • Adds supported_datasets property

arcgis.widgets

• Adds capability to view Tiled Imagery Layers in ArcGIS Online with the map widget

Fixes

Guides

• Get Started
  • Install and set up
    • Updates to Table of Contents
    • Updates links to Deep Learning installer and instructions
    • Adds link to Deep Learning Essentials as alternative for installing Deep Learning Frameworks
• Deep Learning with ArcGIS
  • [Inference only Text Models in arcgis.learn](https://developers.arcgis.com/python/guide/inference-only-text-models-in-arcgis-learn/)
• Working with the Spatially Enabled DataFrame
  • Spelling and grammar edits to Introduction to the Spatially Enabled DataFrame

Samples

• GIS analysts and data scientists
  • Batch size reduced in ChangeDetector samples to avoid cuda errors
  • Connection edits in Land Parcel Extraction using Edge Detection model
  • Fixes ValueError errors in Time Series Analysis of the 2019 Novel Coronavirus Pandemic

Module Changes

arcgis.gis

• GIS
  • Fixes response handling when GIS object initialized with either the api_key or client_id parameter
• Item
  • Fixes error with tables property on items sourced from cached Map Services
  • Fixes share() method when item is part of a Hub and item is in a Group and members can edit the item
  • Fixes BUG-000136169 share() method fails to share item to shared update group when run by non-owner admin
  • Fixes download() failures for csv files output from ArcGIS Online usage reports
  • Fixes usage() issues if input Feature Service is empty or contains only Tables
• ContentManager
  • Fixes failures when trying to use clone_items() on items with optimized layers
• UserManager
  • Fixes search() method return values to populate the provider property

arcgis.features

• FeatureLayer
  • Fixes issue with query(as_df=True) failing if FeatureLayer has null geometries
• Spatially Enabled DataFrame
  • Fixes to_featureset() failure if DataFrame contains a DateTime field with all NaT values
  • Fixes issue with object created using the from_layer() method on a layer with undefined Geometry column (Table) was not passing an empty renderer

arcgis.features.analysis

• arcgis.features.use_proximity
  • Fixes run failures with measurement_type=StraightLine on:
    • connect_origins_to_destinations()
    • find_nearest()
• arcgis.features.summarize_data()
  • Fixes summarize_within() to make sum_within_layer optional so bin_type, bin_size, and bin_size_unit can be used

arcgis.raster

• ImageryLayer
  • Fixes identify() to accept a dictionary as the geometry argument
  • Fixes issue when colormap() was not used for tiles only services
  • Fixes Unable to complete operation error when from_geometry argument is a dictionary
  • Removes internal properties from documentation
• Raster
  • Fixes failurs with applying or calling operations on ArcPy engine Raster objects with raster functions applied
• RasterCollection
  • Fixes issue initializing an object in environments with ArcPyif all the input rasters point to the same Raster object
  • Fixes issue initializing an object from local mosaic datasets where output object is missing Raster key
  • Fixes select_bands() error AttributeError: _'RasterCollection' object has no attribute 'ras_coll_engine_obj'
  • Fixes issues with mosaic() method
    • default value FIRST not set if not value is provided
    • output is not for object created from image services
  • Fixes issues with applying or calling operations on individual raster objects from a RasterCollection created from an ImageryLayer that was the output of create_image_collection() in ArcGIS Online

arcgis.raster.analytics

• Fixes failures caused by missing response information in job results when analyzing ArcGIS Online content
• create_image_collection()
  • Fixes issues when data to upload exceeds 2GB
  • Fixes GIS('pro') or GIS('home') authentication while logged into ArcGIS Online and creating hosted imagery from local raster datasets
• copy_raster()
  • Fixes GIS('pro') or GIS('home') authentication while logged into ArcGIS Online and creating hosted imagery from local raster datasets
• manage_multidimensional_raster()
  • Fixes NameError when using the input_multidimensional_rasters parameter

arcgis.geoanalytics

• analyze_patterns
  • Fixes AttributeError for find_point_clusters() from improperly formatted request
  • Fixes issue with forest() method not handling response results correctly
• find_locations
  • Fixes geocode_locations to properly send the country parameter
• manage_data
  • Fixes issue with run_python_script() tool not properly passing code argument
• summarize_data
  • Fixes error with aggregate_points() when using bins as the aggregation type

arcgis.geocoding

• Fixes issue with properties returned from ArcGIS Enterprise configuration when accessed from Notebook Server
• Modifies internal workings of batch_geocode() to handle large tables

arcgis.learn

• Fixes FileExistsErrors when training models (fit()) with MultiGPU machines
• Data Preparation Methods
  • Fixes ValueError when prepare_tabulardata() is run with a string as the variable_predict argument
• Object Classification Models
  • Fixes AttributeError: _is_multispectral when using FeatureClassifier(backend='tensorflow') and a DataBunch created outside of the ArcGIS Platform
• Feature, Tabular and TimeSeries Models
  • MLModel
    • Fixes missing publish parameter in save() method
    • Fixes errors with show_results() and data.show_batch() when used with dataframe obtained using train_test_split from sklearn
    • Fixes error with predict() not publishing if prediction_type=raster
• PixelClassfication
  • Fixes AttributeError with segmentation models' improper handling of fastai databunches
  • Fixes edge detection models' inferencing output that does not display output raster without changing symbology to Classify if the training data has discontinuous classes
  • ConnectNet
    • Fixes failure to return the mean IOU when loading from_model() without data parameter
  • MultiTaskRoadExtractor
    • Removes UserWarning when using linknet architecture to create model
• Object Detection Models
  • Fixes predict() failures when databunch is prepared with resize_to parameter
  • Fixes training failures with the resnet18 and resnet34 backbones:
    • MaskRCNN
    • FasterRCNN
  • YOLOv3
    • Fixes issue when using Detect Objects Using Deep Learning tool in ArcGIS Pro on model trained with ArcGIS API for Python
  • FasterRCNN
    • Fixes AttributeError failures when running fit(tensorboard=True)
    • Fixes RuntimeError when using save as tflite if data is kitti type in object detection
• Object Tracking Models
  • SiamMask
    • Fixes errors when running import arcgis caused by internal import statements
    • Fixes save() method when object initialized without data
    • Fixes issue when running prepare_data() on Linux
• 3D Models
  • PointCNN
    • Fixes remapping issue in the predict_h5() method when used with output from prepare_data() with remap_classes and background_classcodes arguments, and then exported by the API
• Image Translation Models
• Fixes incorrect output with rows parameter in show_results() when using the hourglass architecture:
  • CycleGAN
  • Pix2Pix
• CycleGAN
  • Fixes incorrect output issues when:
    • calling fit() directly after lr_find()
    • calling fit() 2 or more times on the same model
• Pix2Pix
  • Fixes error with fit() method
  • Removes incorrect accuracy parameter from compute_metrics()
• 3D Models
  • Fixes ValueError when running fit(tensorboard=True) because of plotly orca dependency
• Scanned Maps
  • ScannedMapDigitizer
    • Fixes documentation for:
      • color_list parameter in create_mask() method
      • kernel_size parameter in create_template_image() method
      • color parameter in prepare_search_region()