The team is delivering softwares for image analysis and machine learning, with applications in remote sensing.

POT (Python Optimal Transport): POT is an open source Python library that provides several solvers for optimization problems related to Optimal Transport for signal, image processing and machine learning. It has more than 500k downloads and 1300 stars on github. In 2021, we have received a support from CNRS to develop this software in the context of the national AI plan, with two engineers working part time on the development of the toolbox.

tslearn (Python Optimal Transport): tslearn is a general-purpose Python machine learning library for time series that offers tools for pre-processing and feature extraction as well as dedicated models for clustering, classification and regression. It follows scikit-learn‘s Application Programming Interface for transformers and estimators, allowing the use of standard pipelines and model selection tools on top of tslearn objects.

Triskele: TRISKELE stands for Tree Representations of Images for Scalable Knowledge Extraction and Learning for Earth observation. Triskele is an open source C++ library that provides several algorithms for building hierarchical representation of remote sensing images. It also includes usefull fonctionnalities to produce sobel or NDVI layers and Pantex index as well. (CeCILL-B licence)

Broceliande: Broceliande is a software for classification of remote sensing images. It uses TRISKELE and Random Forests. This software is used in industrial environnements to produce land cover mapping drive by EU projects. (CeCILL-B licence).

Korrigan: Korrigan is a software to search patches in remote sensing databases based on Pattern Spectra.The goal is to offer data mining on big remote sensing image data bases. (CeCILL-B licence)

  • Source Code:

Deep Learning for Remote Sensing: pretrained SegNet architecture.

From Nicolas Audebert (PhD student with ONERA/DTIM and IRISA/OBELIX):
Code repository (GitHub) and
CaffeeModel for ISPRS Vaihingen dataset (IR-R-G image).

UTP (see our BIDS 2016 paper for more details).


Uncompress the archive. It contains the C++ code to copy trees (BPT, alpha-tree, …) from C++ to C++/Java/Python.

It does not include the building algorithm tree, so can’t compile the sources.

GraphBPT (see our ISMM 2015 paper for more details).


Uncompress the archive then click on it or use the following command:

java -jar IsmmBpt2015.jar [–help] [-D dataDir] [image]

Github repository

TreeGen (see the GBR 2015 paper for more details).

Contact the author to get access to the tool

Dense Bag-of-Temporal-SIFT-Words (D-BoTSW) for Time Series Classification (see the LNAI 2016 paper for more details).


For more details, please contact the team members.

Comments are closed.