Version 3.5.5

Major:

• Add backward compatibility loading checkpoint

Minor:

• Change TEST.POST_PROCESSING.MEASURE_PROPERTIES.REMOVE_BY_PROPERTIES.STAT to TEST.POST_PROCESSING.MEASURE_PROPERTIES.REMOVE_BY_PROPERTIES.STATS
• Only check lr scheduler when train in enabled

Bugs fixed:

• Fix a bug in DATA.FILTER_BY_IMAGE
• Update 3D_cell_detection_zarr_tutorial.yaml with new configuration

Full Changelog: v3.5.4...v3.5.5

Version 3.5.4

Major:

• Add BMZ exportation through configuration

Minor:

• Set automatically BMZ path and change it to PATHS.BMZ_EXPORT_PATH

Bugs fixed:

• Fix minor bug when loading model checkpoint
• Fix small bug in semantic seg. multiclass jaccard calculation

Full Changelog: v3.5.3...v3.5.4

Version 3.5.3

Major:

• Update BMZ model check to support more models and increase it's robustness.

Minor:

• Add class extraction for semantic seg. BMZ models.
• Adapt instance segmentation channels to a default value depending when loading BMZ models.
• Change LOAD_MODEL_FROM_CHECKPOINT default value to True.
• Increase UNETR building process robustness

Bugs fixed:

• Fix bug when filtering by entire images.
• Prevent top-5-accuracy when classes are less than 5 in classification workflow.
• Fix bug in single data generator used in classification and SSL workflows.
• Allow BMZ/Torchvision models override completely configuration with the variables they are imposing by making update_dependencies() config function more generic.
• Force entire image filtering when DATA.EXTRACT_RANDOM_PATCH is enabled.

Full Changelog: v3.5.2...v3.5.3

Version 3.5.2

Major:

• Add 'resunet_se' to I2I workflow
• Extend BMZ model support
• Remove DATA.TRAIN.MINIMUM_FOREGROUND_PER. Now for training, validation and test a sample filtering can be made with DATA.TRAIN.FILTER_SAMPLES, DATA.VAL.FILTER_SAMPLES and DATA.TEST.FILTER_SAMPLES respectively.

Minor:

• Add MODEL.LOAD_MODEL_FROM_CHECKPOINT variable
• DATA.PREPROCESS.MEDIAN_BLUR.FOOTPRINT changed to DATA.PREPROCESS.MEDIAN_BLUR.KERNEL_SIZE
• Add robust semantic mask check using DATA.*.CHECK_DATA
• Divide BMZ model check into two functions so they can be reused easily by the GUI
• Change BMZ COLLECTION_URL to a new version of it

Bugs fixed:

• Fix bug in detection workflow when predicting with Zarr/H5 by chunks
• Minor fix during instance training data creation using Zarr
• Correct minor errors during BMZ model import/export

Full Changelog: v3.5.1...v3.5.2

Version 3.5.1

Major:

• Add GRN, ConvNeXtBlock_V2 and UpConvNeXtBlock_V2 blocks

Minor:

• Change CENTRAL_POINT_DILATION from int to list
• Add support for fixed_zero_mean_unit_variance preprocessing for BMZ models
• Adapt BMZ model check function to work properly with models in 0.4 and 0.5 version

Bugs fixed:

• Bug in instance segmentation using only C channel
• Correct tags in BMZ model creation
• Adapt BMZ model check function to work properly with models in 0.4 and 0.5 version

Full Changelog: v3.5.0...v3.5.1

Version 3.5.0

Major

• Move from Pytorch 2.2.0 to 2.4.0
• Increase loss options
• Add options to choose train/test metrics to measure (TRAIN.METRICS/TEST.METRICS). Also added more metrics to measure in Super-resolution, Image to Image and Self-supervised workflows. Closes #86
• Allow central point to use an ellipse footprint in Detection workflow
• Add U-NeXt V1 model
• Update BMZ code to import, finetune and export v0_4 and v0_5 spec models. Move to bioimageio.core version 0.6.7
• Avoid loading entire data when creating instance labels
• Add TEST.DET_IGNORE_POINTS_OUTSIDE_BOX option for Detection workflow
• Remove data clipping during DA so the user is aware of the transformations
• Add LOSS.CLASS_REBALANCE option
• Update all notebooks with more visualization cells and descriptions.

Minor

• Remove TEST.EVALUATE option
• Organize better semantic segmentation output
• Add zoom as preprocessing in test data (only available TEST.BY_CHUNKS == True)
• Reorganize median filter post-processing by creating TEST.POST_PROCESSING.MEDIAN_FILTER, TEST.POST_PROCESSING.MEDIAN_FILTER_AXIS and TEST.POST_PROCESSING.MEDIAN_FILTER_SIZE. Removing TEST.POST_PROCESSING.YZ_FILTERING, TEST.POST_PROCESSING.YZ_FILTERING_SIZE, TEST.POST_PROCESSING.Z_FILTERING and TEST.POST_PROCESSING.Z_FILTERING_SIZE
• Go back to use channel 0 as semantic mask to grow the instances in BC channels. Also fix resolution and channel order in edt() call in BP.
• Remove contrast/brightness EM transformations as they were not used
• Be more permissive with provided csv file during point mask creation in detection workflow
• Add -v option to consult BiaPy's version

Fixes

• Fix resunet++ issue. Closes #95
• Number of residual groups in RCAN.
• Correctly read result images when reusing results. Closes #94
• Avoid loading entire data when creating instance labels. Closes #92
• Add load checkpoint after training again (last epoch training model has been using before)

New Contributors

• @anepaniagua made their first contribution in #85

Full Changelog: v3.4.6...v3.5.0

Version 3.4.6

Minor:

• Change BMZ interaction to allow training using BiaPy's normalization
• Add pooch as a dependency
• Add 'scale_range' norm option
• Improve image and it corresponding gt file matching in instance seg. workflow

Full Changelog: v3.4.5...v3.4.6

Version 3.4.5

Major:

• Allow training with BMZ models.

Minor:

• Add SYSTEM.DEVICE option to allow other backends apart from cpu such as macOS chipset

Bugs fixed:

• Fix minor error when predicting multi-channel data using Zarr
• Fix bug in per patch loss/IoU calculation with batch size > 1

Full Changelog: v3.4.4...v3.4.5

Version 3.4.4

Major:

• Add I2I workflow notebooks

Minor:

• Allow I2I workflow to be operative for 3D images
• Add I2I YAML templates
• Add I2I 3D experiment in run_check.py

Fix:

• Correct models' activation check

Full Changelog: v3.4.3...v3.4.4

Version 3.4.3

Minor:

• Improve the messaging of some errors to make them more comprehensible for the end user.
• Restrict TEST.POST_PROCESSING.REPARE_LARGE_BLOBS_SIZE usage to instance segmentation workflow and BP channels.
• Allow detection and denoising workflows use unetr and multiresunet models.
• Change affine AUGMENTOR.AFFINE_MODE to reflect by default.
• Now the grid in the aumented samples saved take into account the image size to alway create 5x5 grid.
• Adapt MAE's grid mask to be operative in 3D.

Bug fixes:

• 3D stack metric values.
• Fix errors in DATA.PROBABILITY_MAP.
• Prevent TEST.POST_PROCESSING.CLEAR_BORDER remove all instances in 2D.
• Fix minor bugs during some of the instance segmentation post-processing due image shape mismatch.
• Now TEST.POST_PROCESSING.CLEAR_BORDER and TEST.POST_PROCESSING.VORONOI_ON_MASK are above TEST.POST_PROCESSING.MEASURE_PROPERTIES.REMOVE_BY_PROPERTIES so the instances can be repaired before filtering and stats.
• Minor bugs in detection watershed
• Fix bug during detection mask creation for Zarr images with more than one channel
• Fix process_sample_by_chunks() function call in multi-gpu setting due to recent changes
• Fix errors in percentile clipping
• Avoid stuck processes to jump into inference phase in multi-gpu configuration when setting patience during training
• Fix cross-validation errors: 1) when using it in SR workflow due to its upsampling ; 2) in classification
• Fix bug in grid masking using mae in SSL workflow
• Add preprocessing function into classification
• Minor bug when C is not the last channel using Zarr inference

Full Changelog: v3.4.2...v3.4.3