Skip to content

Releases: wri/sentinel-tree-cover

v1.1.0

21 Jul 15:36
Compare
Choose a tag to compare

This release includes some major updates, including:

  • Reduced data throughput requirements by using preprocessed cloud probabilities
  • Better removal of clouds and cloud shadows
  • Improved mosaicing and tiling of prediction windows
  • Improved unit test coverage
  • Type hints and docstrings for the majority of code
  • Swap float32 file structure to uint16 for reduced storage demand
  • Swap numpy array saving to hickle to reduce storage demand
  • Improve multitemporal alignment by doing a median correction between interpolated and non-interpolated regions
  • Adds example data to the large-area notebooks
  • Includes beta integration capabilities with the SEPAL platform

v1.0.0

13 May 14:44
bc2c631
Compare
Choose a tag to compare
v1.0.0 Pre-release
Pre-release

Source code for "A global method to identify trees inside and outside of forests with medium-resolution satellite imagery"