- Fixed styling, changed all string from single quotes to double quotes, corrected comments, etc.
- Added
CVImage
andPillowImage
objects inmltu.annotations.image
to handle image annotations
- Added
RandomSaltAndPepper
augmentor intomltu.augmentors
- Added
SelfAttention
layer intomltu.tensorflow.layers
to use with Conv2D layers (need more testings).
- Fixed bug in
mltu.dataProvider.DataProvider
object to work withoutdata_preprocessors
when feeding loaded data in memory
- Added
RandomGaussianBlur
augmentor intomltu.augmentors
- Fix
ImageReader
to work either with image path ornp.ndarray
- Added
metadata
support tocallbacks/tf2onnx
when converting to onnx format
- Changed
mltu.augmentors
to work only withImage
objects
- Created
Image
object inmltu.annotations.image
to handle image annotations
- changes
OnnxInferenceModel
inmltu.torch.inferenceModels
to load custom metadata from saved ONNX model - improved
mltu.dataProvider
to remove bad samples from dataset on epoch end
- added
mltu.torch.losses
, used to create PyTorch losses, that may be used in training and validation - added CTC loss to
mltu.torch.losses
that can be used for training CTC based models - added
Model2onnx
andTensorboard
callbacks tomltu.torch.callbacks
, used to create PyTorch callbacks, that may be used in training and validation - added
CERMetric
andWERMetric
tomltu.torch.metrics
, used to create PyTorch metrics, that may be used in training and validation - created 08 pytorch tutorial, that shows how to use
mltu.torch
to train CTC based models
- In all tutorials removed stow dependency and replaced with os package, to make it easier to use on Windows 11
- added
mltu.torch
, that contains PyTorch utilities for training machine learning models - added
mltu.torch.dataProvider
, used to create PyTorch data loaders for training and validation - added
mltu.torch.models
, used to create PyTorch models, that wrapps whole model pipeline (training, validation, metrics, callbacks, etc.) - added
mltu.torch.callbacks
, used to create PyTorch callbacks, that may be used in training and validation - added
mltu.torch.metrics
, used to create PyTorch metrics, that may be used in training and validation - added
07_pytorch_tutorial
tutorial
- detaching TensorFlow from mltu, now mltu is only a collection of utilities for training machine learning models
- added 06_pytorch_introduction tutorial
- added
mltu.tensorflow
andmltu.torch
into built package
- added 05_sound_to_text tutorial
- added
WavReader
tomltu/preprocessors
, used to read wav files and convert them to numpy arrays
- added
mltu.utils
into built package
- seperated
CWERMetric
toCER
andWER
Metrics inmltu.metrics
, Character/word rate was calculatted in a wrong way - created @setter for augmentors and transformers in DataProvider, to properlly add augmentors and transformers to the pipeline
- augmentors and transformers must inherit from
mltu.augmentors.base.Augmentor
andmltu.transformers.base.Transformer
respectively - updated ImageShowCV2 transformer documentation
- fixed OnnxInferenceModel in
mltu.inferenceModels
to use CPU even if GPU is available with force_cpu=True flag
- added RandomSharpen to mltu.augmentors, used for simple image augmentation;
- added ImageShowCV2 to mltu.transformers, used to show image with cv2 for debugging purposes;
- added better explained documentation
- created unittests for CER and WER in mltu.utils.text_utils and TensorFlow verion of CER and WER mltu.metrics
- added mltu.augmentors (RandomBrightness, RandomRotate, RandomErodeDilate) - used for simple image augmentation;
Initial release of mltu (Machine Learning Training Utilities)
- Project to help with training machine learning models