Releases: ProjectNeura/MIPCandy
Releases · ProjectNeura/MIPCandy
1.1.1
What's Changed
- Shape mismatch in
_SegmentationLossby @ATATC in #208 dice_similarity_coefficient_with_logits(): RuntimeError: Boolean value of Tensor with more than one value is ambiguous by @ATATC in #210DiceBCELossWithLogitsstill usesmin_percentage_per_classby @ATATC in #214- Overlay fails when
num_classis 1 by @ATATC in #216 - Replace
one_hotwithscatter_inconvert_ids_to_logits()speedup by @perctrix in #211 - Remove dice with logits clip by @ATATC in #218
inspect(): batchtolist()and reusefg_mask(#220) by @perctrix in #221convert_ids_to_logits()does not work with single-channel IDs by @ATATC in #223- Support automatic broadcasting in
_Lossby @ATATC in #227 - Extend
Profilerto support memory leak diagnostics by @ATATC in #232 - Update homepage by @ATATC in #234
inspect(): raise error when the image and label have different shapes. by @ATATC in #238
Full Changelog: 1.1.0...1.1.1
1.1.1-beta.1
What's Changed
- Update homepage by @ATATC in #234
inspect(): raise error when the image and label have different shapes. by @ATATC in #238
Full Changelog: 1.1.1-beta.0...1.1.1-beta.1
1.1.1-beta.0
What's Changed
- Support automatic broadcasting in
_Lossby @ATATC in #227 - Extend
Profilerto support memory leak diagnostics by @ATATC in #232
Full Changelog: 1.1.1-alpha.1...1.1.1-beta.0
1.1.1-alpha.1
What's Changed
Full Changelog: 1.1.1-alpha.0...1.1.1-alpha.1
1.1.1-alpha.0
What's Changed
- Shape mismatch in
_SegmentationLossby @ATATC in #208 dice_similarity_coefficient_with_logits(): RuntimeError: Boolean value of Tensor with more than one value is ambiguous by @ATATC in #210DiceBCELossWithLogitsstill usesmin_percentage_per_classby @ATATC in #214- Overlay fails when
num_classis 1 by @ATATC in #216 - Replace
one_hotwithscatter_inconvert_ids_to_logits()speedup by @perctrix in #211 - Remove dice with logits clip by @ATATC in #218
inspect(): batchtolist()and reusefg_mask(#220) by @perctrix in #221
Full Changelog: 1.1.0...1.1.1-alpha.0
1.1.0
What's Changed
- Support loading a checkpoint in
Trainerby @ATATC in #108 - Multiple values for
selfinTrainer.save_everything_for_recovery()by @ATATC in #112 - Incorrect unpacking order of paddings in Restore3d causes dimension mismatch by @perctrix in #115
TrainerTracker.worst_caseinvolvestorch.Tensorthat causesasdict()failure by @ATATC in #116- Support optionally saving and loading toolbox by @ATATC in #117
- Add random and foreground oversampling in
ROIDatasetby @perctrix in #83 SegmentationTrainervisualization suppresses small values by @ATATC in #121soft_dice_coefficientincorrectly flattens all classes together by @perctrix in #125- Finish README by @ATATC in #124
- Make images on the home page zoomable by @ATATC in #128
- Mismatched headings on the home page by @ATATC in #130
- Inapproapriate use of space in home page navigation bar by @ATATC in #132
SlidingTrainer.validate_case()should compute metrics on full volume after sliding window reconstruction by @perctrix in #139SegmentationTrainer.save_preview()should check the number of label channels by @ATATC in #145SegmentationTrainer.save_preview()should check the number of output channels by @ATATC in #147- For consistency, introduce types
Shape2dandShape3dby @ATATC in #148 - Fix incorrect
_min_factorindex mapping inPad3d.forward()by @perctrix in #150 - Fix multiple bugs in
inspection.pyby @perctrix in #152 - Missing export for
WandBFrontendby @ATATC in #154 Pad2dandPad3d: mismatched padding order by @ATATC in #157Normalize: vanished gradient due to in-place operations by @ATATC in #159SegmentationTrainershould generate preview based on class ids not logits by @ATATC in #161- Missing dependencies
pyyaml,torchvision, andwandbby @ATATC in #163 - Incorrect Fold Picking Logic by @ATATC in #164
- Sanity check inside
Trainer.train()will OOM before sliding window by @ATATC in #166 - Fix
SlidingTrainer.get_example_input()to return correct kernel size by @perctrix in #172 - Rename
include_bgtoinclude_backgroundby @perctrix in #169 ROIDatasetis not foldable by @ATATC in #175- Add transform support in all datasets by @ATATC in #176
- Support joining transforms by @ATATC in #183
- Fix "Support joining transforms" by @ATATC in #184
save_image()can never save RGB images by @ATATC in #186load_image():ensure_num_dimensions()should append dims at the end by @ATATC in #188- Sliding window: fix batch processing logic by @ATATC in #180
- Support
torch.compile()by @ATATC in #189 - Empty tensors result in Nan visualizations by @ATATC in #195
- Preprocess sliding window by @ATATC in #193
auto_convert()has unnecessary overhead when the image is empty by @ATATC in #196- Fixed stride inconsistency between
do_sliding_windowandrevert_sliding_windowby @perctrix in #198 - Beat nnU-Net by @ATATC in #168
SupervisedDataset.fold()does not pass the desired types of arguments toconstruct_new()by @ATATC in #203- Support dice on logits by @ATATC in #206
Full Changelog: 1.0.2...1.1.0
1.1.0-beta.3
1.1.0-beta.2
What's Changed
SegmentationTrainer.save_preview()should check the number of output channels by @ATATC in #147- For consistency, introduce types
Shape2dandShape3dby @ATATC in #148
Full Changelog: 1.1.0-beta.1...1.1.0-beta.2
1.1.0-beta.1
What's Changed
SlidingTrainer.validate_case()should compute metrics on full volume after sliding window reconstruction by @perctrix in #139SegmentationTrainer.save_preview()should check the number of label channels by @ATATC in #145
Full Changelog: 1.1.0-beta.0...1.1.0-beta.1