Creates training iterator to access and augment the dataset

tefla.core.iter_ops.create_training_iters (cnf, data_set, standardizer, crop_size, epoch, parallel=True)

Args

  • cnf: configs dict with all training and augmentation params
  • data_set: an instance of the dataset class
  • standardizer: data samples standardization; either samplewise or aggregate
  • crop_size: training time crop_size of the data samples
  • epoch: the current epoch number; used for data balancing
  • parallel: iterator type; either parallel or queued

Creates prediction iterator to access and augment the dataset

tefla.core.iter_ops.create_prediction_iter (cnf, standardizer, crop_size, preprocessor=None, sync=False)

Args

  • cnf: configs dict with all training and augmentation params
  • standardizer: data samples standardization; either samplewise or aggregate
  • crop_size: training time crop_size of the data samples
  • preprocessor: data processing or cropping function
  • sync: a bool, if False, used parallel iterator