Version 1.1.7
Major:
- Update BiaPy-core code to
3.5.11release - Reorganize the code and add typing
- Change how the BMZ models are checked to avoid loading all the RDF files and speed up the proccess
- Options moved to general tab:
PATCH_SIZE- All configuration for the models
- Some data configuration:
DATA.NORMALIZATION,TRAIN.BATCH_SIZE,DATA.REFLECT_TO_COMPLETE_SHAPEandDATA.FORCE_RGB
Minor:
- Hide
IGNORE_POINTS_OUTSIDE_BOXin the config file when it is not set - Add new BMZ fields
- Change default batch size to 1 to avoid errors
- Change Zarr/H5 related vars from
BY_CHUNKSsetting to the test data - Log error when the container crashed trying to run
Fixes:
- Fix bugs in detection workflow variables due to recent changes
- Ensure batch size is a positive int
- Fix
DATA.*.FILTER_SAMPLESvariable conversion - Fix problems with some models and add more models to workflows
- Reset pretrained model fields when changing workflow or dimensions
- Ensure nickname display in BMZ models as, for example, in Linux, they were not displayed correctly
Full Changelog: v1.1.6...v1.1.7