v2024.1 (January 25, 2024)

Sequential Quantification and Classification modeling:  Apply a unique quantification model to each class in sequence after a classification model. This can provide higher accuracy for quantification analysis in relevant use cases.

Model performance statistics: Model table and Obs vs. calc plot now includes RMSE, RPD and R² also for training, test and CV for comprehensive model comparison (for CV when using full cross validation).

Statistics for Quantification results: Add absolute, relative, and percentage error metrics to the Statistics (properties) descriptor, see Statistics (properties)

View spectral band model statistics: Show spectral plots for model loadings, weights, importance, and SNR in the model wizard

Custom Python Interpreter Selection: Added option to select a custom Python interpreter, offering greater flexibility when training a custom prediction model or using any Python function.

• Python script (descriptor)

• Python script (segmentation)

• Training ONNX model using external python script file

Cross-validation settings panel in model wizard: Select full- or partial-cross validation, number of CV groups and exclusion protocol.

Data Slicing Descriptor: Divide the output of calculated contiuous properties into classes based on specific intervals. Data slicing

Pseudo-RGB Image Descriptor: Combine any 3 properties to generate insightful pseudo-RGB images. Combined RGB image

Summary visualization for "number of pixels by given class": Bar chart visualization in the Table view showing the number of pixels by given class for  each object.

Zoom Overview in Image: Enhances image examination by showing the position of the zoom area in the full image.

Pixel coordinate segmentation preview: Preview image shows segmented area for the measurement using the entered pixel coordinates.

Set max and min values for single bands image: Option to change the scale for the single band intensity image in Pixel Explore.

Labels for selected spectra: Write names and select colors for selected and pinned spectra in spectral plots in Explore and Pixel Explore.

Create class category from Identifier and Group name: Use sample Identifier column or group name to generate a class category column. 

Evince shortcut for object level exploration: Open the object level data in Evince through the Explore tab in Breeze for expanded visualization possibilities.

Selection av samples in Table using classes: Select samples in Table based on descriptor classes.

Spectral plot colored by class in Model wizard: Spectral data is colored by class for each spectrum, making it easier to see which bands are different for different classes.

Pixel Explore Upgrades: More options in Pixel Explore, including changing the visualization of your spectra to raw data, reflectance or absorbance, view wavelength names of RGB bands, and a new zoom toolbar.

View model train/test info in table: Display of a train/test column in the model descriptors table. Train / Test

Compatibility with more camera manufacturers and models

• INNO-SPEC: Speccer (stage), RedEye1.7 HighRes, ColEx (RGB+HSI). INNO-SPEC

• HAIP Solutions: BlackIndustry VNIR and SWIR, HAIP Solutions

Max length for sample mover: Set a max value for better control of stage movement.

Multiple camera connection: Improved support for connecting multiple cameras simultaneously (up to 10 supported).

Camera setting using frame period or frames per second:  Option to use frame period or fps in Hardware settings.

Mineral group classification: USGS PRISM MICA can now also be used for classification of your data into mineral groups in addition to individual minerals. The mineral groups are shown in the Select end members dialog for easy selection of relevant materials groups.

Compartments for depth registration: Simplifies the adding of compartments in the Depth registration Panel. Simply add number of rows, start position, number of compartments, length of each compartment and gap between them. Then use the mouse to adjust the position.