Version: [3241] Whitebox Workflows for QGIS 2.1.3

## 2.1.3 - 2026-06-14

### Fixed
- Fixed classifier tools showing a file-picker text field for the
class label field parameter instead of a dropdown selector. Affected
tools include `knn_classification`, `logistic_regression`,
`random_forest_classification`, `fuzzy_knn_classification`,
`nnd_classification`, `min_dist_classification`, and
`parallelepiped_classification`. The parameter schema for
`class_field` (and similar attribute field parameters) was incorrectly
typed as a LiDAR file input, causing the QGIS plugin to render a file
picker. Additionally, the parameter type-inference pre-pass was using
only heuristic inference rather than explicit schema kinds, causing
`training_data` vectors to be misclassified. Both issues are now
resolved: schema kinds take precedence over heuristics, and attribute
field parameters are correctly promoted to
`QgsProcessingParameterField` dropdowns. Users can now select from
available fields in the training layer without manual text entry.
- Fixed `obia_pipeline_basic` and `classify_objects_random_forest`,
`classify_objects_svm`, and `classify_objects_ensemble_pro` OBIA tools
incorrectly treating `training` and `class_field` parameters as LiDAR
inputs. These tools now correctly render `training` as a CSV/Table
file picker and `class_field` as a field name text input.
- Added curated parameter descriptions for
`random_forest_classification_fit` and `random_forest_regression_fit`
tools in the QGIS parameter metadata, providing users with clear
guidance on band ordering, scaling options, and model usage.

yes

jlindsay

2026-06-14T17:11:24.767872+00:00

3.28.0

4.99.0

None

no

Version management

Plugin details