Changelog

v0.7.2

  • Minor bug fixing.

v0.7.1

  • In “Band classification” tool added feature importance output when using Random Forest with scikit framework.

  • Minor bug fixing.

v0.7.0

  • New optional dependency mpi4py for MPI bindings.

  • Implementation of MPI in several tools for parallel processing.

v0.6.3

  • Fix band name for several products in download_products by Cemu0.

v0.6.2

  • Minor bug fixing.

v0.6.0

  • Added optional dependency Pandas for performance improvement in tabular data.

  • In tool “Band classification” added option for using PyTorch pretrained model. In case a pretrained model is selected, and additional algorithm is selected for classification, using the same parameters of the named algorithm (e.g. random forest); after executing the pretrained model, the additional algorithm is executed on the embeddings for classification. Currently, it works with models pretrained by the Allen Institute for Artificial Intelligence (SatlasPretrain: https://satlas-pretrain.allen.ai) in particular, Sentinel-2 swin-v2-base single-image multispectral and swin-v2-tiny single-image multispectral models, and Landsat 8 Landsat 9 swin-v2-base single-image multispectral model. SatlasPretrain model weights are released under the Open Data Commons ‘Attribution License (ODC-BY). The repository code is licensed under the Apache License 2.0 (https://huggingface.co/allenai/satlas-pretrain). This tool downloads the official SatlasPretrain weights (Bastani et al., “SatlasPretrain: A Large-Scale Dataset for Remote Sensing Image Understanding”, ICCV 2023, arXiv:2211.15660, https://doi.org/10.48550/arXiv.2211.15660). All model weights remain the property of their respective authors.

  • In tool “Band classification” added PyTorch pretreained segmentation models for Sentinel-2 (swin-v2-base single-image multispectral model using 3 bands or 4 bands) pretrained by DPR Team as part of the DPR Zoo Segmentation Hub framework (https://github.com/DPR25/dpr-zoo-segmentation-hub) based on SatlasPretrain models. The model output classes: background, water, developed, tree, shrub, grass, crop, bare, snow, wetland, mangroves, moss. The repository code of DPR Zoo models are licensed under the MIT License (https://huggingface.co/martinkorelic/dpr-zoo-models). This tool downloads the model weights (DPR Team, 2025. Made as part of Arnes Hackathon 2025). All model weights remain the property of their respective authors.

  • In tool “Download products” included the download of Sentinel-2 L2A SCL band.

  • In tool “Preprocess products” included the Sentinel-2 L2A SCL band.

  • Improved progress monitoring for multiprocess.

  • Code optimization and bug fixing.

v0.5.2

  • Performance improvement for the tool “Vector to raster” with the method area_based.

  • Improvement of the multiprocess iterator.

  • Minor fixes

v0.5.1

  • Fixed issue with the tool “Vector to raster” where a few polygons were randomly skipped using the method area_based.

  • Minor fixes

v0.5.0

  • New tool “Raster label” for calculating the area of contiguous patches in a raster. The output is a raster where each pixel value represents the pixel count of the patch thereof.

  • Performance improvement for the tool “Vector to raster” with the method area_based.

  • Minor fixes

v0.4.4

  • Changed raster_zonal_stats to accept raster input as reference_path

  • Fixed handling nan value as nodata

v0.4.3

  • First experimental implementation of Pytorch for band_calc

  • Minor fixes

v0.4.2

  • Minor fixes

v0.4.1

  • Fixed preprocessing calculation

  • Minor fixes

v0.4.0

  • Added tool “Band clustering” for unsupervised K-means classification of bandset

  • Added tool “Raster edit” for direct editing of pixel values based on vector

  • Added tool “Raster zonal stats” for calculating statistics of a raster intersecting a vector.

  • Improved the NoData handling for multiprocess calculation

  • In “Band clip”, “Band dilation”, “Band erosion”, “Band sieve”, “Band neighbor”, “Band resample” added the option multiple_resolution to keep original resolution of individual rasters, or use the resolution of the first raster for all the bands

  • In “Cross classification” fixed area based accuracy and added kappa hat metric

  • In “Band combination” added option no_raster_output to avoid the creation of output raster, producing only the table of combinations

  • In “Band calc” replaced nanpercentile with optimized calculation function

  • Improved extraction of ROIs in “Band classification”

  • Minor bug fixing and removed Requests dependency

v0.3.5

  • Fixed Copernicus access token error

  • Fixed automatic band wavelength definition in BandSet

v0.3.04

  • Fixed Jupyter interface

v0.3.03

  • Fixed Jupyter interface

v0.3.02

  • Fixed Jupyter interface

v0.3.01

  • Added functions for interactive interface in Jupyter environment

  • Fixed Sentinel-2 band 8A identification in preprocess products

v0.2.01

  • In Download Products added the functions to search and download Collections from Microsoft Planetary Computer: Sentinel-2, Landsat, ASTER, MODIS Surface Reflectance 8-Day, and Copernicus DEM

v0.1.24

  • Fixed band calc calculation with multiband raster as bandset

  • Fixed preview path for Copernicus products

v0.1.23

  • Minor fixes

v0.1.22

  • Fixed prepare input function

  • Fixed logger for multiprocess

v0.1.21

  • Fixed requirements

v0.1.20

  • Fixed Copernicus search and download service

v0.1.19

  • Fixed Copernicus search and download service

v0.1.18

v0.1.17

  • Fixed spectral signature calculation for multiband raster

  • Fixed closing multiprocess at exit

v0.1.16

  • Fixed issue in block size calculation for multiprocess in case of large input raster and low RAM;

  • Fixed management of bandsets using multiband rasters;

  • Minor fixes to multiprocess download;