v2022.2 (October 24, 2022)
Modelling and algorithms
Train models using Python script Training ONNX model using external python script file
Model log and version history
Optimization metric for Machine learning training can be selected
Minimum Wavelength Mapping (MWL) BREEZE GEO MWL mapping
USGS PRISM MICA with import of MICA color file. BREEZE GEO USGS PRISM MICA classification
Spatial values descriptor in either mm, m, feet or inches Shape and size
Image segmentation
Shape file image segmentation Shapefile
Dimension (number of pixels) for each selection when using representative spectrum
White reference extraction from image
Hardware
Unispectral Monarch II camera supported for data acquisition Unispectral
Hyspex dual camera data acquisition with merging and aligning of image data with Key point detection (SURF, SIFT and ORB) and RANSAC, and detector jump adjustments
Record length can be set in seconds
Data import and export
Envi classification format (use with for example QGIS), is available as a format for exporting classifications in image data.
Data ignore value in Envi files is handled
YOLOv5 annotation format for export of spectral images
Visualization and interaction
Histogram for properties and predicted variables in Table
More colors options for classification categories
Pixel tooltip for mouse pointer in legend panel for measurement image
Spectral filters can selected in Pixel Explore
Pretreatments steps can be shown in Analyse Tree, more on Pretreatments
Rename study or models in overview
Invert selection in plots and images
Copy comments to clipboard
BREEZE GEO Requires separate license