[R-sig-Geo] Release of SITS version 1.0.0

[GilbertoCamara]

Dear R-SIG-GEO list members,

I would like to inform you that the “sits” open source R package version 1.0.0 has been accepted in CRAN. 

“sits” is a package for satellite image time series analysis using machine learning that works with big image collections available in cloud services. Users can perform the full cycle of land user and land cover classification: 

1. Select an image collection available on cloud providers AWS, Microsoft Planetary Computer, Digital Earth Africa and Brazil Data Cube. 
2. Build a regular data cube using analysis-ready image collections.
3. Extract labelled time series from data cubes to be used as training samples.
4. Perform quality control using self-organised maps.
5. Filtering time series samples for noise reduction.
6. Use the samples to train machine learning models.
7. Tune machine learning models for improved accuracy.
8. Classify data cubes using machine learning models.
9. Post-process classified images with Bayesian smoothing to remove outliers. 
10. Estimate uncertainty values of classified images.
11. Evaluate classification accuracy using best practices.
12. Improve results with active learning and self-supervised learning methods.

The package is equivalent to Google Earth Engine for land classification, plus some functions not available in the standard GEE API (sample quality analysis using SOM, time series deep learning methods, model tuning, uncertainty measures, active learning and self-supervised learning). The library has been developed since 2016 and has now reached TRL 8 status, being currently used in Brazil for classification of the Amazonia and Cerrado biomes.

Our work relies on many important contributions for the R-SIG-GEO community. Many thanks to Edzer Pebesma for “sf” and “stars”, Robert Hijmans for “terra”, Marius Appel for “gdalcubes”, Tim Appelhans for “leafem”, Victor Maus for “dtwSat”. We are also indebted to the RStudio team for the “tidyverse”, to Daniel Falbel for “torch” and “luz”, and to Dirk Eddelbuettel for “Rcpp” and “RcppArmadillo”. A big round of appreciation goes to Roger Bivand, who put us the Right path 20 years ago. 

“Sits” can be found on GitHub at "https://github.com/e-sensing/sits” and documentation in "https://e-sensing.github.io/sitsbook/“. 

Best regards
Gilberto
============================
Prof Dr Gilberto Camara
Senior Researcher
National Institute for Space Research (INPE), Brazil
https://gilbertocamara.org/
=============================