Satellite Image Time Series Analysis for Earth Observation Data
Cubes
Description
An end-to-end toolkit for land use and land cover classification
using big Earth observation data, based on machine learning methods
applied to satellite image data cubes, as described in Simoes et al (2021) .
Builds regular data cubes from collections in AWS, Microsoft Planetary Computer,
Brazil Data Cube, Copernicus Data Space Environment (CDSE), Digital Earth Africa, Digital Earth Australia,
NASA HLS using the Spatio-temporal Asset Catalog (STAC)
protocol () and the 'gdalcubes' R package
developed by Appel and Pebesma (2019) .
Supports visualization methods for images and time series and
smoothing filters for dealing with noisy time series.
Includes functions for quality assessment of training samples using self-organized maps
as presented by Santos et al (2021) .
Provides machine learning methods including support vector machines,
random forests, extreme gradient boosting, multi-layer perceptrons,
temporal convolutional neural networks proposed by Pelletier et al (2019) ,
and temporal attention encoders by Garnot and Landrieu (2020) .
Supports GPU processing of deep learning models using torch .
Performs efficient classification of big Earth observation data cubes and includes
functions for post-classification smoothing based on Bayesian inference, and
methods for active learning and uncertainty assessment. Supports object-based
time series analysis using package supercells .
Enables best practices for estimating area and assessing accuracy of land change as
recommended by Olofsson et al (2014) .
Minimum recommended requirements: 16 GB RAM and 4 CPU dual-core.