Version 1.2.0-4

Hotfix version 1.2.0-4

• Fix issue #918

Version 1.2.0-3

Hotfix version 1.2.0-3

• Fix stars proxy bug (issue #902)
• Fix purrr cross deprecation
• Fix ggplot2 aes_string deprecation

Version 1.2.0-1

Hotfix version 1.2.0-1

• Fix sits_som_clean_samples() bug (issue #890)

Version 1.2.0

New features in SITS version 1.2.0

• sits_get_data() can be used to retrieve samples in classified cube
• Support for mixture models (sits_mixture_model())
• Joining cubes in a mosaic (sits_mosaic_cubes())
• Extract the trained ML model (sits_model())
• Downloading and copying data cubes (sits_cube_copy())
• Combine prediction by average and entropy (sits_combine_predictions())
• Significant performance improvement when working with COG files
• Allow plot of confusion matrix (sits_plot)
• Support for operations on CLOUD band in sits_apply()
• Bug fixes and internal re-engineering for better code maintenance

Version 1.1.0

What's changed in SITS version 1.1.0

• Introduced support to kernel functions in sits_apply
• Introduced new function sits_mixture_model for spectral mixture analysis
• Support for the Swiss Data Cube (swissdatacube.org)
• Support for mosaic visualization in sits_view
• Introduced new function sits_as_sf to convert sits objects to sf
• Export images as COG in sits_regularize
• Add roi parameter in sits_regularize function
• Add crs parameter in sits_get_data
• Change Microsoft Planetary Computer source name to "MPC"
• Fix several bugs and improve performance

Version 1.0.0

Available on CRAN
Changes to match the package to CRAN standards
Update documentation

Version 0.17.0

What's changed

• Introduced new classifier model sits_lighttae() (Lightweight Temporal Self-Attention)
• Introduced sits_uncertainty_sampling() for active learning
• Introduced sits_confidence_samples() for semi-supervised learning
• Introduced sits_geo_dist() to generate samples-samples and samples-predicted plot
• Introduced sits_tuning() for random search of machine learning parameters
• Introduced sits_reduce_imbalance() function to balance class samples
• Introduced sits_as_sf() to convert a sits tibble to a sf object
• Support to torchopt deep learning optimizer package
• New types of sits_uncertainty(): least confidence and margin of confidence
• Implement parallel processing for sits_kfold_validate()
• Change data to samples in sits machine learning classifiers (NOTE: models trained in previous versions is no longer supported)
• Change deep learning functions to snake case
• Remove file parameter in sits_get_data() function
• Update documentation
• Improve several internal functions performances
• Fix several bugs

Version 0.16.3

What's changed

• reimplemented all deep learning functions using torch package and remove keras dependence
• Introduced sits_TAE() classification model
• Introduced sits_lightgbm() classification model
• Simplified sits_regularize() parameters
• Improve sits_regularize() to reach production level quality
• Improve sits_regularize() to use C++ internal functions
• Include improved version of gdalcubes
• Improve sits_cube() to open results cube
• Update plot() parameters on raster cubes
• Support multi-tile for classified cube in sits_view()

Version 0.16.2

What's Changed

• Improve sits_get_data() to accept tibbles
• Remove multiples progress bar from sits_cube()
• Improve sits_regularize() to process in parallel by tiles, bands, and dates
• Improve sits_regularize() to check malformed files

Version 0.16.1

What's Changed

• fix AWS environment variables
• add resume feature and fault tolerance in sits_apply() function