Thomas Oberlin
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Thomas Oberlin has a total of 84 co-authors. Here are the top co-authors by number of shared publications:
- Nicolas Dobigeon
IRIT/INP-ENSEEIHT, University of Toulouse, Toulouse, France
10 shared publications - Cédric Févotte
UMR 7373 Aix-Marseille Université CNRS I2M Marseille France
7 shared publications Sylvain Meignen
7 shared publications- Paul Magron
Université de Lorraine CNRS Inria LORIA F-54000 Nancy France
5 shared publications - Clovis Tauber
UMR Inserm U930, Université François-Rabelais de Tours, Tours, France
4 shared publications - Dennis G. Wilson
ISAE-SUPAERO, University of Toulouse
4 shared publications
Latest Publications
- 9 May 2026
- Published by Springer Nature
- p. 501-517
Multi-objective Optimization for Synthetic-to-Real Style Transfer
- 9 May 2026
- Published by Springer Nature
- p. 501-517
Multi-objective Optimization for Synthetic-to-Real Style Transfer
- 31 March 2026
- Published by Agence Bibliographique de l'Enseignement Supérieur
Analyse de signaux multicomposantes : contributions à la décomposition modale Empirique, aux représentations temps-fréquence et au Synchrosqueezing
Les superpositions d'ondes modulées en amplitude et en fréquence (modes AM--FM) sont couramment utilisées pour modéliser de nombreux signaux du monde réel : cela inclut des signaux audio (musique, parole), médicaux (ECG), ou diverses séries temporelles (températures, consommation électrique). L'objectif de ce travail est l'analyse et la compréhension fine de tels signaux, dits "multicomposantes" car ils contiennent plusieurs modes. Les méthodes mises en oeuvre vont permettre de les représenter efficacement, d'identifier les différents modes puis de les démoduler (c'est-à-dire déterminer leur amplitude et fréquence instantanée), et enfin de les reconstruire. On se place pour cela dans le cadre bien établi de l'analyse temps-fréquence (avec la transformée de Fourier à court terme) ou temps-échelle (transformée en ondelettes continue). On s'intéressera également à une méthode plus algorithmique et moins fondée mathématiquement, basée sur la notion de symétrie des enveloppes des modes : la décomposition modale empirique. La première contribution de la thèse propose une alternative au processus dit ``de tamisage'' dans la décomposition modale empirique, dont la convergence et la stabilité ne sont pas garanties. \`A la place, une étape d'optimisation sous contraintes ainsi qu'une meilleure détection des extrema locaux du mode haute fréquence garantissent l'existence mathématique du mode, tout en donnant de bons résultats empiriques. La deuxième contribution concerne l'analyse des signaux multicomposantes par la transformée de Fourier à court terme et à la transformée en ondelettes continues, en exploitant leur structure particulière ``en ridge'' dans le plan temps-fréquence. Plus précisément, nous proposons une nouvelle méthode de reconstruction des modes par intégration locale, adaptée à la modulation fréquentielle, avec des garanties théoriques. Cette technique donne lieu à une nouvelle méthode de débruitage des signaux multicomposantes. La troisième contribution concerne l'amélioration de la qualité de la représentation au moyen de la ``réallocation'' et du ``synchrosqueezing''. Nous prolongeons le synchrosqueezing à la transformée de Fourier à court terme, et en proposons deux extensions inversibles et adaptées à des modulations fréquentielles importantes, que nous comparons aux méthodes originelles. Une généralisation du synchrosqueezing à la dimension 2 est enfin proposée, qui utilise le cadre de la transformée en ondelettes monogène.
- 31 March 2026
- Published by Agence Bibliographique de l'Enseignement Supérieur
Analyse de signaux multicomposantes : contributions à la décomposition modale Empirique, aux représentations temps-fréquence et au Synchrosqueezing
Les superpositions d'ondes modulées en amplitude et en fréquence (modes AM--FM) sont couramment utilisées pour modéliser de nombreux signaux du monde réel : cela inclut des signaux audio (musique, parole), médicaux (ECG), ou diverses séries temporelles (températures, consommation électrique). L'objectif de ce travail est l'analyse et la compréhension fine de tels signaux, dits "multicomposantes" car ils contiennent plusieurs modes. Les méthodes mises en oeuvre vont permettre de les représenter efficacement, d'identifier les différents modes puis de les démoduler (c'est-à-dire déterminer leur amplitude et fréquence instantanée), et enfin de les reconstruire. On se place pour cela dans le cadre bien établi de l'analyse temps-fréquence (avec la transformée de Fourier à court terme) ou temps-échelle (transformée en ondelettes continue). On s'intéressera également à une méthode plus algorithmique et moins fondée mathématiquement, basée sur la notion de symétrie des enveloppes des modes : la décomposition modale empirique. La première contribution de la thèse propose une alternative au processus dit ``de tamisage'' dans la décomposition modale empirique, dont la convergence et la stabilité ne sont pas garanties. \`A la place, une étape d'optimisation sous contraintes ainsi qu'une meilleure détection des extrema locaux du mode haute fréquence garantissent l'existence mathématique du mode, tout en donnant de bons résultats empiriques. La deuxième contribution concerne l'analyse des signaux multicomposantes par la transformée de Fourier à court terme et à la transformée en ondelettes continues, en exploitant leur structure particulière ``en ridge'' dans le plan temps-fréquence. Plus précisément, nous proposons une nouvelle méthode de reconstruction des modes par intégration locale, adaptée à la modulation fréquentielle, avec des garanties théoriques. Cette technique donne lieu à une nouvelle méthode de débruitage des signaux multicomposantes. La troisième contribution concerne l'amélioration de la qualité de la représentation au moyen de la ``réallocation'' et du ``synchrosqueezing''. Nous prolongeons le synchrosqueezing à la transformée de Fourier à court terme, et en proposons deux extensions inversibles et adaptées à des modulations fréquentielles importantes, que nous comparons aux méthodes originelles. Une généralisation du synchrosqueezing à la dimension 2 est enfin proposée, qui utilise le cadre de la transformée en ondelettes monogène.
- Open Access
Multi-Objective Optimization for Synthetic-to-Real Style Transfer
Semantic segmentation networks require large amounts of pixel-level annotated data, which are costly to obtain for real-world images. Computer graphics engines can generate synthetic images alongside their ground-truth annotations. However, models trained on such images can perform poorly on real images due to the domain gap between real and synthetic images. Style transfer methods can reduce this difference by applying a realistic style to synthetic images. Choosing effective data transformations and their sequence is difficult due to the large combinatorial search space of style transfer operators. Using multi-objective genetic algorithms, we optimize pipelines to balance structural coherence and style similarity to target domains. We study the use of paired-image metrics on individual image samples during evolution to enable rapid pipeline evaluation, as opposed to standard distributional metrics that require the generation of many images. After optimization, we evaluate the resulting Pareto front using distributional metrics and segmentation performance. We apply this approach to standard datasets in synthetic-to-real domain adaptation: from the video game GTA5 to real image datasets Cityscapes and ACDC, focusing on adverse conditions. Results demonstrate that evolutionary algorithms can propose diverse augmentation pipelines adapted to different objectives. The contribution of this work is the formulation of style transfer as a sequencing problem suitable for evolutionary optimization and the study of efficient metrics that enable feasible search in this space. The source code is available at: https://github.com/echigot/MOOSS.
Open AccessMulti-Objective Optimization for Synthetic-to-Real Style Transfer
Semantic segmentation networks require large amounts of pixel-level annotated data, which are costly to obtain for real-world images. Computer graphics engines can generate synthetic images alongside their ground-truth annotations. However, models trained on such images can perform poorly on real images due to the domain gap between real and synthetic images. Style transfer methods can reduce this difference by applying a realistic style to synthetic images. Choosing effective data transformations and their sequence is difficult due to the large combinatorial search space of style transfer operators. Using multi-objective genetic algorithms, we optimize pipelines to balance structural coherence and style similarity to target domains. We study the use of paired-image metrics on individual image samples during evolution to enable rapid pipeline evaluation, as opposed to standard distributional metrics that require the generation of many images. After optimization, we evaluate the resulting Pareto front using distributional metrics and segmentation performance. We apply this approach to standard datasets in synthetic-to-real domain adaptation: from the video game GTA5 to real image datasets Cityscapes and ACDC, focusing on adverse conditions. Results demonstrate that evolutionary algorithms can propose diverse augmentation pipelines adapted to different objectives. The contribution of this work is the formulation of style transfer as a sequencing problem suitable for evolutionary optimization and the study of efficient metrics that enable feasible search in this space. The source code is available at: https://github.com/echigot/MOOSS.
- 18 November 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Geoscience and Remote Sensing
- Vol. 63 (01962892) , 1-16
Deep Priors for Satellite Image Restoration With Accurate Uncertainties
Satellite optical images, upon their on-ground receipt, offer a distorted view of the observed scene. Their restoration, including denoising, deblurring, and sometimes super-resolution, is required before their exploitation. Moreover, quantifying the uncertainties related to this restoration helps to reduce the risks of misinterpreting the image content. Deep learning methods are now state-of-the-art for satellite image restoration. Among them, direct inversion methods train a specific network for each sensor, and generally provide a point estimation of the restored image without the associated uncertainties. Alternatively, deep regularization (DR) methods learn a deep prior on target images before plugging it, as the regularization term, into a model-based optimization scheme. This allows for restoring images from several sensors with a single network and possibly for estimating associated uncertainties. In this paper, we introduce VBLE-xz, a DR method that solves the inverse problem in the latent space of a variational compressive autoencoder (CAE). We adapt the regularization strength by modulating the bitrate of the trained CAE with a training-free approach. Then, VBLE-xz estimates relevant uncertainties jointly in the latent and in the image spaces by sampling an explicit posterior estimated within variational inference. This enables fast posterior sampling, unlike state-of-the-art DR methods that use Markov chains or diffusion-based approaches. We conduct a comprehensive set of experiments on very high-resolution simulated and real Pléiades images, asserting the performance, robustness and scalability of the proposed method. They demonstrate that VBLE-xz represents a compelling alternative to direct inversion methods when uncertainty quantification is required.
- 18 November 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Geoscience and Remote Sensing
- Vol. 63 (01962892) , 1-16
Deep Priors for Satellite Image Restoration With Accurate Uncertainties
Satellite optical images, upon their on-ground receipt, offer a distorted view of the observed scene. Their restoration, including denoising, deblurring, and sometimes super-resolution, is required before their exploitation. Moreover, quantifying the uncertainties related to this restoration helps to reduce the risks of misinterpreting the image content. Deep learning methods are now state-of-the-art for satellite image restoration. Among them, direct inversion methods train a specific network for each sensor, and generally provide a point estimation of the restored image without the associated uncertainties. Alternatively, deep regularization (DR) methods learn a deep prior on target images before plugging it, as the regularization term, into a model-based optimization scheme. This allows for restoring images from several sensors with a single network and possibly for estimating associated uncertainties. In this paper, we introduce VBLE-xz, a DR method that solves the inverse problem in the latent space of a variational compressive autoencoder (CAE). We adapt the regularization strength by modulating the bitrate of the trained CAE with a training-free approach. Then, VBLE-xz estimates relevant uncertainties jointly in the latent and in the image spaces by sampling an explicit posterior estimated within variational inference. This enables fast posterior sampling, unlike state-of-the-art DR methods that use Markov chains or diffusion-based approaches. We conduct a comprehensive set of experiments on very high-resolution simulated and real Pléiades images, asserting the performance, robustness and scalability of the proposed method. They demonstrate that VBLE-xz represents a compelling alternative to direct inversion methods when uncertainty quantification is required.
- Open Access
Synthetic Data for Robust Runway Detection
Deep vision models are now mature enough to be integrated in industrial and possibly critical applications such as autonomous navigation. Yet, data collection and labeling to train such models requires too much efforts and costs for a single company or product. This drawback is more significant in critical applications, where training data must include all possible conditions including rare scenarios. In this perspective, generating synthetic images is an appealing solution, since it allows a cheap yet reliable covering of all the conditions and environments, if the impact of the synthetic-to-real distribution shift is mitigated. In this article, we consider the case of runway detection that is a critical part in autonomous landing systems developed by aircraft manufacturers. We propose an image generation approach based on a commercial flight simulator that complements a few annotated real images. By controlling the image generation and the integration of real and synthetic data, we show that standard object detection models can achieve accurate prediction. We also evaluate their robustness with respect to adverse conditions, in our case nighttime images, that were not represented in the real data, and show the interest of using a customized domain adaptation strategy.
Open AccessSynthetic Data for Robust Runway Detection
Deep vision models are now mature enough to be integrated in industrial and possibly critical applications such as autonomous navigation. Yet, data collection and labeling to train such models requires too much efforts and costs for a single company or product. This drawback is more significant in critical applications, where training data must include all possible conditions including rare scenarios. In this perspective, generating synthetic images is an appealing solution, since it allows a cheap yet reliable covering of all the conditions and environments, if the impact of the synthetic-to-real distribution shift is mitigated. In this article, we consider the case of runway detection that is a critical part in autonomous landing systems developed by aircraft manufacturers. We propose an image generation approach based on a commercial flight simulator that complements a few annotated real images. By controlling the image generation and the integration of real and synthetic data, we show that standard object detection models can achieve accurate prediction. We also evaluate their robustness with respect to adverse conditions, in our case nighttime images, that were not represented in the real data, and show the interest of using a customized domain adaptation strategy.
- 18 September 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Signal Processing Letters
- Vol. 32 (10709908) , 3705-3709
Online Simplex-Structured Matrix Factorization
Simplex-structured matrix factorization (SSMF) is a common task encountered in signal processing and machine learning. Minimum-volume constrained unmixing (MVCU) algorithms are among the most widely used methods to perform this task. While MVCU algorithms generally perform well in an offline setting, their direct application to online scenarios suffers from scalability limitations due to memory and computational demands. To overcome these limitations, this paper proposes an approach which can build upon any off-the-shelf MVCU algorithm to operate sequentially, i.e., to handle one observation at a time. The key idea of the proposed method consists in updating the solution of MVCU only when necessary, guided by an online check of the corresponding optimization problem constraints. It only stores and processes observations identified as informative with respect to the geometrical constraints underlying SSMF. We demonstrate the effectiveness of the approach when analyzing synthetic and real datasets, showing that it achieves estimation accuracy comparable to the offline MVCU method upon which it relies, while significantly reducing the computational cost.
- 18 September 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Signal Processing Letters
- Vol. 32 (10709908) , 3705-3709
Online Simplex-Structured Matrix Factorization
Simplex-structured matrix factorization (SSMF) is a common task encountered in signal processing and machine learning. Minimum-volume constrained unmixing (MVCU) algorithms are among the most widely used methods to perform this task. While MVCU algorithms generally perform well in an offline setting, their direct application to online scenarios suffers from scalability limitations due to memory and computational demands. To overcome these limitations, this paper proposes an approach which can build upon any off-the-shelf MVCU algorithm to operate sequentially, i.e., to handle one observation at a time. The key idea of the proposed method consists in updating the solution of MVCU only when necessary, guided by an online check of the corresponding optimization problem constraints. It only stores and processes observations identified as informative with respect to the geometrical constraints underlying SSMF. We demonstrate the effectiveness of the approach when analyzing synthetic and real datasets, showing that it achieves estimation accuracy comparable to the offline MVCU method upon which it relies, while significantly reducing the computational cost.
- 17 September 2025
- Published by Springer Nature
- p. 294-304
Synthetic Data for Robust Runway Detection
- Open Access
Online simplex-structured matrix factorization
Simplex-structured matrix factorization (SSMF) is a common task encountered in signal processing and machine learning. Minimum-volume constrained unmixing (MVCU) algorithms are among the most widely used methods to perform this task. While MVCU algorithms generally perform well in an offline setting, their direct application to online scenarios suffers from scalability limitations due to memory and computational demands. To overcome these limitations, this paper proposes an approach which can build upon any off-the-shelf MVCU algorithm to operate sequentially, i.e., to handle one observation at a time. The key idea of the proposed method consists in updating the solution of MVCU only when necessary, guided by an online check of the corresponding optimization problem constraints. It only stores and processes observations identified as informative with respect to the geometrical constraints underlying SSMF. We demonstrate the effectiveness of the approach when analyzing synthetic and real datasets, showing that it achieves estimation accuracy comparable to the offline MVCU method upon which it relies, while significantly reducing the computational cost.
Open AccessOnline simplex-structured matrix factorization
Simplex-structured matrix factorization (SSMF) is a common task encountered in signal processing and machine learning. Minimum-volume constrained unmixing (MVCU) algorithms are among the most widely used methods to perform this task. While MVCU algorithms generally perform well in an offline setting, their direct application to online scenarios suffers from scalability limitations due to memory and computational demands. To overcome these limitations, this paper proposes an approach which can build upon any off-the-shelf MVCU algorithm to operate sequentially, i.e., to handle one observation at a time. The key idea of the proposed method consists in updating the solution of MVCU only when necessary, guided by an online check of the corresponding optimization problem constraints. It only stores and processes observations identified as informative with respect to the geometrical constraints underlying SSMF. We demonstrate the effectiveness of the approach when analyzing synthetic and real datasets, showing that it achieves estimation accuracy comparable to the offline MVCU method upon which it relies, while significantly reducing the computational cost.
- Open Access
- 15 July 2025
- Published by Elsevier in Science of Remote Sensing
- Vol. 12
Early-season delineation of agricultural fields using a fully convolutional multi-task network and satellite images
- 15 July 2025
- Published by Elsevier in Science of Remote Sensing
- Vol. 12
Open AccessEarly-season delineation of agricultural fields using a fully convolutional multi-task network and satellite images
- 2 May 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Image Processing
- Vol. 34 (10577149) , 2896-2909
Variational Bayes Image Restoration With Compressive Autoencoders
Regularization of inverse problems is of paramount importance in computational imaging. The ability of neural networks to learn efficient image representations has been recently exploited to design powerful data-driven regularizers. While state-of-the-art plug-and-play (PnP) methods rely on an implicit regularization provided by neural denoisers, alternative Bayesian approaches consider Maximum A Posteriori (MAP) estimation in the latent space of a generative model, thus with an explicit regularization. However, state-of-the-art deep generative models require a huge amount of training data compared to denoisers. Besides, their complexity hampers the optimization involved in latent MAP derivation. In this work, we first propose to use compressive autoencoders instead. These networks, which can be seen as variational autoencoders with a flexible latent prior, are smaller and easier to train than state-of-the-art generative models. As a second contribution, we introduce the Variational Bayes Latent Estimation (VBLE) algorithm, which performs latent estimation within the framework of variational inference. Thanks to a simple yet efficient parameterization of the variational posterior, VBLE allows for fast and easy (approximate) posterior sampling. Experimental results on image datasets BSD and FFHQ demonstrate that VBLE reaches similar performance as state-of-the-art PnP methods, while being able to quantify uncertainties significantly faster than other existing posterior sampling techniques. The code associated to this paper is available in https://github.com/MaudBqrd/VBLE.
- 2 May 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Image Processing
- Vol. 34 (10577149) , 2896-2909
Variational Bayes Image Restoration With Compressive Autoencoders
Regularization of inverse problems is of paramount importance in computational imaging. The ability of neural networks to learn efficient image representations has been recently exploited to design powerful data-driven regularizers. While state-of-the-art plug-and-play (PnP) methods rely on an implicit regularization provided by neural denoisers, alternative Bayesian approaches consider Maximum A Posteriori (MAP) estimation in the latent space of a generative model, thus with an explicit regularization. However, state-of-the-art deep generative models require a huge amount of training data compared to denoisers. Besides, their complexity hampers the optimization involved in latent MAP derivation. In this work, we first propose to use compressive autoencoders instead. These networks, which can be seen as variational autoencoders with a flexible latent prior, are smaller and easier to train than state-of-the-art generative models. As a second contribution, we introduce the Variational Bayes Latent Estimation (VBLE) algorithm, which performs latent estimation within the framework of variational inference. Thanks to a simple yet efficient parameterization of the variational posterior, VBLE allows for fast and easy (approximate) posterior sampling. Experimental results on image datasets BSD and FFHQ demonstrate that VBLE reaches similar performance as state-of-the-art PnP methods, while being able to quantify uncertainties significantly faster than other existing posterior sampling techniques. The code associated to this paper is available in https://github.com/MaudBqrd/VBLE.