Project

Back to overview

Understanding the underlying structure of remote sensing images: improving adaptation in classification models with artificial intelligence techniques

English title Understanding the underlying structure of remote sensing images: improving adaptation in classification models with artificial intelligence techniques
Applicant Tuia Devis
Number 136827
Funding scheme Ambizione
Research institution Laboratoire de systèmes d'information géographique EPFL - ENAC - IIE - LASIG
Institution of higher education EPF Lausanne - EPFL
Main discipline Other disciplines of Environmental Sciences
Start/End 01.12.2011 - 30.11.2014
Approved amount 322'845.00
Show all

All Disciplines (2)

Discipline
Other disciplines of Environmental Sciences
Other disciplines of Engineering Sciences

Keywords (6)

Remote sensing; Machine learning; Multitemporal image analysis; Statistical models; Disaster managment; Landscape genetics

Lay Summary (English)

Lead
Lay summary

Optical satellite sensors now provide images of high quality, but of increasing complexity. They can provide information with a spatial resolution of less than a meter, cover the region of visible and near infrared spectrum with a hundred narrow spectral bands and revisit the same region at intervals of only a few days. These images, once treated, provide crucial information to scientists of many different domains such as disaster managment or planning of urban policies.

While these domains all increasingly rely on remote sensing imagery, they are so far impeded by technical limitations. This project aims at pushing those limitations back.

The increase in image resolutions (spatial, spectral and temporal) and the sheer number of sensors acquiring images make it impossible to develop a specific model for each image. On the other hand, applying a model developed for one image to another gives poor results, because the changes in illumination, geometry and landuse types on the new image make the data distribution change, or shift.

This project studies the nature of those shift and aims at characterizing its nonlinearities and studying the variations of image structure when acquisition conditions change. Focus is put on statistical modelling of data structure (manifold learning) and on models facilitating transfer of models across data acquisitions (transfer learning). Understanding the changes occurring in manifolds will enhance the potential for models transferability and to surpass the “one image / one model” limitation of current remote sensing data processing research. With the developments of the project, it will be possible to develop adaptable classification models that can process images of different zones, taken at different times and by different sensors, thus filling this major gap between current remote sensing research and end-users expectations.

To ensure this last point, the project also aims at developing validated applicative tools for applications needing landuse maps or environmental parameters retrieved from remote sensing data. To this end, real case studies in landscape genetics, disaster management and atmospheric modelling will be considered.

Summarizing, the project will participate in scientific advances in the fields of machine learning and fill theoretical gaps in current remote sensing image processing research that prevent the field to meet users' expectations.


Direct link to Lay Summary Last update: 21.02.2013

Responsible applicant and co-applicants

Employees

Publications

Publication
Advances in Hyperspectral Image Classification
Camps-Valls Gustavo, Tuia Devis, Bruzzone Lorenzo, Benediktsson Jon Atli (2014), Advances in Hyperspectral Image Classification, in IEEE SIGNAL PROCESSING MAGAZINE, 31(1), 45-54.
Hydrometeor classification from two-dimensional video disdrometer data
Grazioli Jacopo, Tuia Devis, Monhart S., Schneebeli Marc, Raupach Timothy H., Berne Alexis (2014), Hydrometeor classification from two-dimensional video disdrometer data, in Atmospheric Measurement Techniques, 7(9), 2869-2882.
PRINCIPAL POLYNOMIAL ANALYSIS
Laparra Valero, Jiménez Sandra, Tuia Devis, Camps-Valls Gustau (2014), PRINCIPAL POLYNOMIAL ANALYSIS, in International Journal of Neural Systems, 24(7), 1440007.
Semisupervised classification of remote sensing images with hierarchical spatial similarity
Huo Lianzhi, Tang Ping, Zhang Zheng, Tuia Devis (2014), Semisupervised classification of remote sensing images with hierarchical spatial similarity, in IEEE Geoscience and Remote Sensing Letters, 12(1), 150-154.
Semi-supervised multiview embedding for hyperspectral data classification
Volpi Michele, Matasci Giona, Kanevski Mikhail F., Tuia Devis (2014), Semi-supervised multiview embedding for hyperspectral data classification, in Neurocomputing, 145, 427-437.
SVM active learning approach for image classification using spatial information
Pasolli Edoardo, Melgani Farid, Tuia Devis, Pacifici Fabio, Emery William J. (2014), SVM active learning approach for image classification using spatial information, in IEEE Transactions on Geoscience and Remote Sensing, 52(4), 2217-2223.
Active Learning: Any Value for Classification of Remotely Sensed Data?
Crawford Melba M., Tuia Devis, Yang Hsiuhan Lexie (2013), Active Learning: Any Value for Classification of Remotely Sensed Data?, in PROCEEDINGS OF THE IEEE, 101(3), 593-608.
Automatic feature learning for spatio-spectral image classification with sparse SVM
Tuia Devis, Volpi Michele, Mura Mauro Dalla, Rakotomamonjy Alain, Flamary Rémi (2013), Automatic feature learning for spatio-spectral image classification with sparse SVM, in IEEE Transactions on Geoscience and Remote Sensing, 52(10), 6062-6074.
Explicit recursive and adaptive filtering in reproducing kernel hilbert spaces
Tuia Devis, Muñoz-Marí Jordi, Rojo-Álvarez Jose´ Luis, Martínez-Ramón Manel A., Camps-Valls Gustavo (2013), Explicit recursive and adaptive filtering in reproducing kernel hilbert spaces, in IEEE Transactions on Neural Networks and Learning Systems, 25(7), 1413-1419.
Foreword to the special issue on machine learning for remote sensing data processing
Tuia Devis, Merényi Erzsébet, Jia Xiuping, Grana-Romay Manuel (2013), Foreword to the special issue on machine learning for remote sensing data processing, in IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 7(4), 1007-1011.
Graph Matching for Adaptation in Remote Sensing
Tuia Devis, Munoz-Mari Jordi, Gomez-Chova Luis, Malo Jesus (2013), Graph Matching for Adaptation in Remote Sensing, in IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING, 51(1), 329-341.
Learning User's Confidence for Active Learning
Tuia Devis, Munoz-Mari Jordi (2013), Learning User's Confidence for Active Learning, in IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING, 51(2), 872-880.
Semisupervised manifold alignment of multimodal remote sensing images
Tuia Devis, Volpi Michele, Trolliet Maxime, Camps-Valls Gustavo (2013), Semisupervised manifold alignment of multimodal remote sensing images, in IEEE Transactions on Geoscience and Remote Sensing, 52(12), 7708-7720.
Semi-Supervised Novelty Detection Using SVM Entire Solution Path
de Morsier Frank, Tuia Devis, Borgeaud Maurice, Gass Volker, Thiran Jean-Philippe (2013), Semi-Supervised Novelty Detection Using SVM Entire Solution Path, in IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING, 51(4), 1939-1950.
Supervised change detection in VHR images using contextual information and support vector machines
Volpi M, Tuia D, Bovolo F, Kanevski M, Bruzzone L (2013), Supervised change detection in VHR images using contextual information and support vector machines, in INTERNATIONAL JOURNAL OF APPLIED EARTH OBSERVATION AND GEOINFORMATION, 20, 77-85.
Large Margin Filtering
Flamary R, Tuia D, Labbe B, Camps-Valls G, Rakotomamonjy A (2012), Large Margin Filtering, in IEEE TRANSACTIONS ON SIGNAL PROCESSING, 60(2), 648-659.
Memory-Based Cluster Sampling for Remote Sensing Image Classification
Volpi M, Tuia D, Kanevski M (2012), Memory-Based Cluster Sampling for Remote Sensing Image Classification, in IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING, 50(8), 3096-3106.
Multi-Modal Change Detection, Application to the Detection of Flooded Areas: Outcome of the 2009-2010 Data Fusion Contest
Longbotham N, Pacifici F, Glenn T, Zare A, Volpi M, Tuia D, Christophe E, Michel J, Inglada J, Chanussot J, Du Q (2012), Multi-Modal Change Detection, Application to the Detection of Flooded Areas: Outcome of the 2009-2010 Data Fusion Contest, in IEEE JOURNAL OF SELECTED TOPICS IN APPLIED EARTH OBSERVATIONS AND REMOTE SENSING, 5(1), 331-342.
Multiscale analysis of geomorphological and geological features in high resolution digital elevation models using the wavelet transform
Kalbermatten M, Van De Ville D, Turberg P, Tuia D, Joost S (2012), Multiscale analysis of geomorphological and geological features in high resolution digital elevation models using the wavelet transform, in GEOMORPHOLOGY, 138(1), 352-363.
Remote sensing image segmentation by active queries
Tuia D, Munoz-Mari J, Camps-Valls G (2012), Remote sensing image segmentation by active queries, in PATTERN RECOGNITION, 45(6), 2180-2192.
Semisupervised Classification of Remote Sensing Images With Active Queries
Munoz-Mari J, Tuia D, Camps-Valls G (2012), Semisupervised Classification of Remote Sensing Images With Active Queries, in IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING, 50(10), 3751-3763.
SVM-Based Boosting of Active Learning Strategies for Efficient Domain Adaptation
Matasci G, Tuia D, Kanevski M (2012), SVM-Based Boosting of Active Learning Strategies for Efficient Domain Adaptation, in IEEE JOURNAL OF SELECTED TOPICS IN APPLIED EARTH OBSERVATIONS AND REMOTE SENSING, 5(5), 1335-1343.
Unsupervised Change Detection With Kernels
Volpi M, Tuia D, Camps-Valls G, Kanevski M (2012), Unsupervised Change Detection With Kernels, in IEEE GEOSCIENCE AND REMOTE SENSING LETTERS, 9(6), 1026-1030.
Cluster validity measure and merging system for hierarchical clustering considering outliers
De Morsier Frank, Tuia Devis, Borgeaud Maurice, Gass Volker, Thiran Jean Philippe, Cluster validity measure and merging system for hierarchical clustering considering outliers, in Pattern Recognition.

Collaboration

Group / person Country
Types of collaboration
Laboratoire d'Informatique, du Traitement de l'Information et des Systèmes - University of Rouen France (Europe)
- in-depth/constructive exchanges on approaches, methods or results
- Publication
Aerospace Engineering Laboratory - Univeristy of Colorado at Boulder United States of America (North America)
- Publication
Hyperspectral computing laboratory - University of Extremadura Spain (Europe)
- in-depth/constructive exchanges on approaches, methods or results
- Exchange of personnel
Vision lab, university of Antwerpen Belgium (Europe)
- in-depth/constructive exchanges on approaches, methods or results
- Publication
- Exchange of personnel
DLR, Dept. Georisk and Civil Security Germany (Europe)
- in-depth/constructive exchanges on approaches, methods or results
- Publication
- Exchange of personnel
GipsaLab - INP Grenoble France (Europe)
- in-depth/constructive exchanges on approaches, methods or results
- Publication
Université of Nice Sophia Antipolis France (Europe)
- in-depth/constructive exchanges on approaches, methods or results
- Publication
Remote Sensing Laboratory - University of Trento Italy (Europe)
- in-depth/constructive exchanges on approaches, methods or results
- Publication
UNOSAT - UNITAR Switzerland (Europe)
- Industry/business/other use-inspired collaboration
Digital Globe International United States of America (North America)
- Industry/business/other use-inspired collaboration
Université de Bretagne du Sud France (Europe)
- in-depth/constructive exchanges on approaches, methods or results
- Publication
- Exchange of personnel
Image Processing Laboratory - Universitat de València Spain (Europe)
- in-depth/constructive exchanges on approaches, methods or results
- Publication
- Exchange of personnel
Laboratory for Applications of Remote Sensing - Purdue University United States of America (North America)
- in-depth/constructive exchanges on approaches, methods or results
- Publication
Instite of Geomatics and Analysis of Risk - University of Lausanne Switzerland (Europe)
- in-depth/constructive exchanges on approaches, methods or results
- Publication
- Research Infrastructure
- Exchange of personnel

Scientific events

Active participation

Title Type of contribution Title of article or contribution Date Place Persons involved
European Conference on Machine Learning (ECML) Talk given at a conference Domain adaptation with regularized optimal transport. 15.09.2014 Nancy, France Tuia Devis;
Workshop ‘Photogrammetry and Computer Vision’ (PCV) Talk given at a conference An active set strategy for multiclass hyperspectral image classification with group-lasso regularization 06.09.2014 ETH, Switzerland Tuia Devis;
Int. Conf. Pattern Recognition (ICPR) Poster Network-based correlated correspondence for unsupervised domain adaptation of hyperspectral satellite images 24.08.2014 Stockholm, Sweden Tuia Devis;
Int. Conf. Pattern Recognition (ICPR) Poster Unsupervised alignment of image manifolds with centrality measures 24.08.2014 Stockholm, Sweden Tuia Devis;
International Geoscience and Remote Sensing Symposium (IGARSS) Talk given at a conference Domain adaptation in remote sensing through cross-image synthesis with dictionaries 14.07.2014 Québec city, Canada Tuia Devis;
International Geoscience and Remote Sensing Symposium (IGARSS) Talk given at a conference Weakly supervised alignment of images with semantic ties 14.07.2014 Québec city, Canada Tuia Devis;
Swiss Machine learning Day (SMLD) Talk given at a conference Kernel sparse subspace clustering 13.11.2013 EPFL Lausanne, Switzerland Tuia Devis;
International Geoscience and Remote Sensing Symposium (IGARSS) Talk given at a conference Create the relevant spatial filterbank in the hyperspectral jungle. 21.07.2013 Melbourne, Australia Tuia Devis;
International Geoscience and Remote Sensing Symposium (IGARSS) Talk given at a conference Statistical assessment of dataset shift in model portability of multi-angle image acquisitions. 21.07.2013 Melbourne, Australia Tuia Devis;
International Geoscience and Remote Sensing Symposium (IGARSS) Talk given at a conference Multisource alignment of image manifolds 21.07.2013 Melbourne, Australia Tuia Devis;
Joint Urban Remote Sensing Event Talk given at a conference Classification of urban multi-angular image sequences by aligning their manifolds 21.04.2013 Sao Paulo, Brazil Tuia Devis;
International Conference on Pattern Recognition Applications and Methods (ICPRAM) Talk given at a conference nvestigating the feature extraction framework for domain adaptation in remote sensing image classification. 15.02.2013 Barcelona, Spain Tuia Devis;
International Conference on Computer Vision in Remote Sensing (CVRS) Talk given at a conference Pose estimation of landscape images using DEM and orthophotos 16.12.2012 Xiamen, China, China Tuia Devis;
American Geoscience Union Meeting (AGU) Talk given at a conference The influence of sea ice extent variability on the Greenland surface mass and energy balance 09.12.2012 San Francisco, USA, United States of America Tuia Devis;
Machine Learning Workshop (MLWS) Talk given at a conference Novelty and change detection in remote sensing images with kernels 19.11.2012 Lausanne, Switzerland, Switzerland Tuia Devis;
International Conference on Pattern Recognition (ICPR) Talk given at a conference Discovering relevant spatial filterbanks for VHR image classification 11.11.2012 Tokyo, Japan, Japan Tuia Devis;
IEEE Machine Learning for Signal Processing Workshop (MLSP) Talk given at a conference Nonlinear data description with principal polynomial analysis. 23.09.2012 Santander, Spain, Spain Tuia Devis;
International Geoscience and Remote Sensing Symposium (IGARSS) Talk given at a conference PUTTING THE USER INTO THE ACTIVE LEARNING LOOP: TOWARDS REALISTIC BUT EFFICIENT PHOTOINTERPRETATION 23.07.2012 Munich, Germany, Germany Tuia Devis;
9th International Geostatistics Congress Talk given at a conference Using active learning for monitoring networks design: the example of wind power plants sites evaluation 11.06.2012 Oslo, Norway, Norway Tuia Devis;
Neural Information Processing Systems (NIPS) Poster Active multiple kernel learning of wind power resources 12.12.2011 Granada, Spain, Spain Tuia Devis;


Self-organised

Title Date Place
Open Source Geospatial Research and Education Symposium (OGRS) 24.10.2012 Yverdon les bains, Switzerland

Knowledge transfer events

Active participation

Title Type of contribution Date Place Persons involved
Journée de veille technologique 19.03.2013 HEIG-VD, Switzerland Tuia Devis;
Réunion du groupe intercantonal LiDAR et Forêts 13.03.2013 Hotel Mirabeau, Switzerland Tuia Devis;


Self-organised

Title Date Place
Journée d'information sur le projet PRESECS: "réponse des herbages jurassiens à la sécheresse" 12.09.2012 La Frétaz, VD, Switzerland

Communication with the public

Communication Title Media Place Year
Media relations: print media, online media A scientific adventure from Lake Geneva to Lake Baikal Mediacom EPFL Western Switzerland International 2013
Media relations: radio, television Collaboration entre la Russie et l'EPFL La Télé Western Switzerland 2013
Media relations: radio, television Sciences: l'EPFL scanne le Léman La Télé Western Switzerland 2013
Media relations: print media, online media Un ULM vole au dessus du Léman 24 heures Western Switzerland 2013
Media relations: radio, television Une équipe de l'EPFL filme le lac Léman en ULM RTS.ch Western Switzerland 2013
Media relations: print media, online media Le changement climatique, une menace pour le fromage d'alpage? Article in the FLASH EPFL Western Switzerland 2012
Media relations: print media, online media L'impact des canicules sous la loupe des experts article de journal, 24h Western Switzerland 2012

Awards

Title Year
Best paper award (honorable mention), for the paper "An active set strategy for multiclass hyperspectral image classification with group-lasso regularization" presented in in the Workshop ‘Photogrammetry and Computer Vision’ (PCV), Zurich, Switzerland. 2014

Associated projects

Number Title Start Funding scheme
150593 Multimodal machine learning for remote sensing information fusion 01.12.2014 SNSF Professorships

Abstract

In this project, I propose to develop theoretical solutions to practical limitations in remote sensing data analysis.I propose to study the underlying structure of high resolution remote sensing data, to characterize its nonlinearities and to study the variations of this structure when acquisition conditions change. This way, it will be possible to develop adaptable classification models that can process images of different zones, taken at different times and by different sensors, thus filling a major gap in current remote sensing research and meeting the endusers expectations. The project aims at creating models that can be applied to several images, thus allowing to surpass an important limitaiton to the use of remote sensing images in real world applications. To ensure this last point, the project also aims at developing validated applicative tools for applications needing landuse maps or environmental parameters retrieved from remote sensing data. To this end, real case studies in landscape genetics, disaster management and atmospheric modeling will be considered.Summarizing, the project will participate in scientific advances in the fields of machine learning and fill theoretical gaps in current remote sensing image processing research that prevent the field to meet users' expectations.
-