What's new in version 1.8.4
New¶
Guides
¶
- Managing ArcGIS Applications
- Deep Learning with ArcGIS
- How ChangeDetection Works?
- Inference only Text Models in
arcgis.learn
- Retraining Windows and Doors Extraction model
- Unsupervised Machine learning using arcgis.learn guide
- YOLOv3 Object Detector
- Faster R-CNN Object Detector
- Edge Detection with
arcgis.learn
- Image Captioning
- How SequenceToSequence Works
Samples
¶
- GIS Analysts and Data Scientists
- Translating Story Map from one language to another using Deep Learning
- Generating rgb imagery from digital surface model using Pix2Pix
- Change Detection of Buildings from Satellite Imagery
- Land Parcel Extraction using Edge Detection model
- Address Standardization and Correction using SequenceToSequence model
Fixes¶
Guides
¶
- Deep Learning with ArcGIS
- Updates guides with
PointRend
enhancement
- Updates guides with
Samples
¶
- GIS Analysts and Data Scientists
- Updates
Try-It-Live
functionality - Updates samples to correct data paths when
.
in paths
- Updates
- Power Users/Developers
- Updates samples that were missing maps in visualization sections
- Visualize monthly changes in Hirakund reservoir using video
- Fixes
OSERROR: cannot open resource
error when definingcreate_movie()
function to visualize monthly changes
- Fixes
Module Changes¶
arcgis.features
¶
FeatureLayer
- Fixes
Token Required
error onquery()
operations if using OAuth Authentication to connect withGIS()
- Fixes
arcgis.widgets
¶
MapView
- Fixes issue with map widget not drawing in ArcGIS Pro 2.7
arcgis.apps
¶
- dashboard
- Fixes
add_row()
to properly readheight
argument - Fixes
add_column()
to properly readwidth
argument - Fixes issue with failures to synchronize maps to widgets
- Fixes
arcgis.learn
¶
- Pixel Classification Models
MultiTaskRoadExtractor
- Fixes issue where model would not run with ArcGIS Pro training tool
ChangeDetector
- Fixes issue using multispectral data where tool does not give any results when inferenced in ArcGIS Pro using Classify Pixels tool
- Fixes failures of
predict(visualize=True)
using multispectral data - Fixes multispectral model training failures at different chip_sizes
- Fixes variations in visualization when using
show_results()
andshow_batch()
with multispectral data
- Object Classfication Models
FeatureClassifier
- Fixes issue where you cannot create object from an ImageNet type dataset
- Tabular and Feature Data Models
TimeSeriesModel
- Fixes failures on model initialization
MLModel
- Fixes issue where overwrites of the original data frame and respective column names occur during data preparation and with
predict()
- Fixes issue where overwrites of the original data frame and respective column names occur during data preparation and with
- Image Translation Models
CycleGAN
- Fixes issue with
prepare_data()
whereimagery_type
parameter was not treated as optional
- Fixes issue with
Pix2Pix
- Fixes issue with
prepare_data()
whereimagery_type
parameter was not treated as optional - Adds missing
accuracy
andshow_progress
parameters tocompute_metrics()
- Fixes issue with