Skip to content

Changelog#

All notable changes to this project will be documented in this file.

The format is based on Keep a Changelog and this project adheres to Semantic Versioning.

[Unreleased]#

Added#

[0.1.6] - 2021-09-06#

Updated#

  • Update EdsrModel to fix no_upsampling bug where self.args was not stored.

[0.1.5] - 2021-08-30#

Added#

  • New model: PhysicssrModel
  • New model: DrnModel
  • New model: HanModel
  • New model: AwsrnModel

[0.1.4] - 2021-08-17#

Added#

  • calculate_mean_std(dataset) to ~super_image.utils.metrics for calculating RGB pixel mean and standard deviation over a dataset.
  • New model: DrlnModel
  • New model: RcanModel
  • New model: MdsrModel
  • New model: CarnModel
  • New model: PanModel

Updated#

  • Update EdsrModel to include no_upsampling option so it can be reused for JiifModel.

[0.1.3] - 2021-07-28#

Added#

  • TrainDataset(dataset) with support for huggingface datasets.
  • augment_five_crop function for use with dataset.map(augment_five_crop, batched=True)

[0.1.2] - 2021-07-26#

Added#

  • Added metrics EvalMetrics().evaluate(model, eval_dataset) class to calculate PSNR and SSIM on a model and evaluation dataset. Accepts EvalDataset(dataset) with huggingface datasets.

Changed#

  • Replaced EvalDataset(dataset) to use huggingface datasets instead of H5 files.
  • Fixed EvalDataset(dataset) to be robust to wrongly sized HR images (not equals to scaled LR image size).