IMAGES team seminars

IMAGES team: Image Modeling, Analysis, GEometry & Synthesis


Seminars of the team

Upcoming seminars:

23/09/2021 at 10h30, room 1D19 (or a zoom link will be available)

Emanuele Dalsasso, As if by magic: self-supervised training of deep despeckling networks with MERLIN

Speckle fluctuations seriously limit the interpretability of synthetic aperture radar (SAR) images. Speckle reduction has thus been the subject of numerous works spanning at least four decades. Techniques based on deep neural networks have recently achieved a new level of performance in terms of SAR image restoration quality.
In this presentation, I will give you an overview of different techniques (pre-trained, supervised, semi-supervised and self-supervised). Then, I will introduce you a new self-supervised strategy based on the separation of the real and imaginary parts of single-look complex SAR images, called MERLIN (coMplex sElf-supeRvised despeckLINg), and show that it offers a straightforward way to train all kinds of deep despeckling networks.

Jonathan Vacher, Flexibly Regularized Mixture Models and Application to Image Segmentation [1]
: Probabilistic finite mixture models are widely used for unsupervised clustering. These models can often be improved by adapting them to the topology of the data. For instance, in order to classify spatially adjacent data points similarly, it is common to introduce a Laplacian constraint on the posterior probability that each data point belongs to a class. Alternatively, the mixing probabilities can be treated as free parameters, while assuming Gauss-Markov or more complex priors to regularize those mixing probabilities. However, these approaches are constrained by the shape of the prior and often lead to complicated or intractable inference. Here, we propose a new parametrization of the Dirichlet distribution to flexibly regularize the mixing probabilities of over-parametrized mixture distributions. Using the Expectation-Maximization algorithm, we show that our approach allows us to define any linear update rule for the mixing probabilities, including spatial smoothing regularization as a special case. We then show that this flexible design can be extended to share class information between multiple mixture models. We apply our algorithm to artificial and natural image segmentation tasks, and we provide quantitative and qualitative comparison of the performance of Gaussian and Student-t mixtures on the Berkeley Segmentation Dataset. We also demonstrate how to propagate class information across the layers of deep convolutional neural networks in a probabilistically optimal way, suggesting a new interpretation for feedback signals in biological visual systems. Our flexible approach can be easily generalized to adapt probabilistic mixture models to arbitrary data topologies.

Previous seminars:

  • February 3, 2pm: 1. Carlo Alberto Barbano (University of Turin) -COVID-19 detection from CXRs & Entangling and Disentangling deep representations for bias correction — 2. Arthur Ouaknine (Télécom Paris) – Scene understanding using raw radar data for autonomous driving.
  • March 10, 2pm: 1. Raphaël Achddou – Synthetic images as a regularity prior for image restoration neural networks — 2. Lara Raad – Image colorization using adversarial learning and semantic information
  • April 14, 2pm: 1. Vicky Kalogeiton (LIX) – Computer graphics and deep learning — 2. Benoît Dufumier – Contrastive Learning with Continuous Proxy Meta-Data for 3D MRI Classification
  • June 2, 10am: Emin Zerman – Quality Assessment and User Interaction for Volumetric Video Technologies
  • June 3, 10am: Amal Dev Parakkat – Interactive sketch processing
  • June 4, 10am: Francesco Banterle – How Can We Make Good Use of High Dynamic Range for Rendering? A Story With and Without Deep Learning
  • March 10, 2pm: 1. Stéphane Lathuiliere – Deformations in deep models for image and video generation — 2. Alasdair Newson – Understanding and organising the latent space of autoencoder
  • October 21: 1. Emanuele Dalsasso  – Apprentissage profond pour l’imagerie SAR: du débruitage à l’interprétation de scène — 2. Antoine Houdard — Transport optimal pour l’apprentissage de modèles génératifs : application à la synthèse de texture par patchs
  • December 7: Welcome day for the new PhD candidates. Poster presentations.
  • December 9: 1. Mengyu Chu – Detail Synthesis for Fluids and Videos with Deep-Learning Algorithms — 2. Matthis Maillard – Knowledge distillation from multi-modal to mono-modal segmentation networks
  • January 29, 1pm: Hong Sun, Prof Wuhan University – Machine-Learning for Classifications of Remote Sensing Imagery
    – Pixel-Object-Scene Classifications of Optical-SAR-PolSAR Data
  • March 20, 2:30pm: Qi Wang, Assistant Professor, Lab for Neural Engineering and Control, Department of Biomedical Engineering, Columbia University – Brain state dependent thalamic information processing and perception
  • May 27: Christophe Lino – Toward Tools Incorporating Knowledge for Virtual Cameras and Lights
  • June 3: Christian Lessig (Magdeburg University, Germany) – Towards Structure Preserving and Adaptive Simulations in Computer Graphics
  • June 3: Adnane Boukhayma (Univ. Oxford, UK) – Data-driven human shape analysis and synthesis
  • June 3: Kiwon Um (Technical University of Munich, Germany) – Recreating and Understanding Nature with Perceivable Data and Deep Learning
  • December 17: Welcome day: Kiwon Um and PhD presentations



  • January 25, 2pm: Stanley Durrleman (ARAMIS) – Building Digital Models of Alzheimer’s Disease Progression
  • March 23, 2pm: Alasdair Newson – Film Grain Synthesis and Autoencoding Geometric Shapes
  • March 27, 10am: Maria Vakalopoulou (Center for Visual Computing laboratory, CentraleSupélec) – Advanced Computer Vision and Machine Learning Techniques for Remote Sensing and Medical Applications
  • April 3, 10am: Betrand Le Saux (ONERA) – Deep Learning for Scene Understanding, from 2D to 3D
  • April 6, 1pm: Adeline Paiement (Swansea University, UK) – Designing computer vision and deep learning methods for scientific image analysis
  • April 12, 10am: Stavros Tsogkas (Computer science department of the University of Toronto) – Learning mid-level representations for computer vision
  • May 31, 3pm: Ramon Pino Pérez (Univ. Los Andes, Meria, Venezuela) – Fusion de croyances et impossibilité
  • June 1, 11am: Wing-Kin (Ken) Ma (Chinese University of Hong Kong) – Hyperspectral Unmixing in Remote Sensing: Learn the Wisdom There and Go Beyond (Machine Learning Included)
  • June 13, 10:30am: Giovanni Sileno – On the problems of interface: conceptual spaces, pertinence, explainability
  • September 5, 2pm: Maxime Fondin – Segmentation et comptage de follicules ovariens en IRM
  • September 21, 12:30am: Bastien Ponchon – Morphological Multi-scale Decompositions and efficient representations with Auto-Encoders
  • September 27, 3pm: Chris Tralie (postdoctoral associate at the Information Initiative at Duke) – Geometric Audiovisual Signal Processing
  • November 15, 2:30pm: Pierre Roussillon – surface and curve matching using normal cycles, and Jean Feydy – optimal transport


  • 2017
  • January 19, 2pm: Josef Sivic, INRIA – Learning visual representations from Internet data – 3pm: Pietro Gori – Learning the structural organization of anatomical shape complexes
  • February 23, 2pm: Valérie Gouet, IGN – About image-based localization
  • April 27, 3pm: Josiane Zerubia, INRIA – Marked Point Processes for Object Detection and Tracking in High Resolution Images: Applications to Remote Sensing and Biology – 4pm: Alasdair Newson – Stochastic Modeling and Realistic Rendering of Film Grain
  • June 1st, 11am: Julien Tierny, LIP6 – Topological data analysis and TTK software
  • June 15th, 2pm: Charles Deledalle, IMB – MuLoG: MUlti-channel LOgarithm with Gaussian denoising – 3pm: Samy Blusseau – he Human vision compared to models of contours and flat surfaces perception
  • October 19, 2pm: Jesus Angulo (Centre de Morphologie Mathématique des Mines ParisTech) – semi-groupes morphologiques sur des espaces métriques et ultramétriques ;
    Samy Blusseau – analyse morphologique d’images de tenseurs.
  • November 23, 2pm: Damien Rohmer (LIX) – Efficient Developable Surface Modeling: From Garment Design to Paper Animation ; Alasdair Newson – How do Autoencoders Encode Geometric Shapes?
  • 2016
  • January 7, 2016 (MAP5, salle du conseil de l’aile Turing) – SMATI seminar – Nicolas Bonneel (3pm) : Transport optimal en informatique graphique – Camille Sutour (4pm) : Vision nocturne numérique
  • February 11, 2016 (C49) – SMATI seminar – Stanley Durrleman (3pm) : Apprentissage de modèles virtuels de la structure cérébrale à partir de données de neuroimagerie – Maxime Daisy
  • February 18, 2016, 3pm, C49 – Olivier Fercoq: Primal-dual coordinate descent
  • April 7, 2016 (C49) – SMATI seminar
    – Gabriel Peyré (3pm) : Transport optimal numérique et applications –
    Blanche Buet (4pm) : Approximation de surfaces par des varifolds discrets
  • March 24, 2016, 3pm, C47 – Isabelle Bloch: Formal concept analysis and mathematical morphology
  • March 10, 2016 (MAP5, salle de conseil de l’aile Turing) – SMATI seminar
    – Julien Rabin (3pm) : Segmentation d’images et transport optimal d’histogrammes – Yann Traonmilin (4pm) : Un cadre pour la reconstruction
    de signaux de faible complexité et son application à la parcimonie structurée

(4pm) : Inpainting basé motif d’images et de vidéos apokiqué aux données stéréoscopiques avec carte de profondeur

  • June 9, 2016, 3pm (B316) – Mariano Tepper: Matrix factorization for big data: From data analysis to ensemble problems
  • June 2, 2016 (C49) – SMATI seminar – François Malgouyres (3pm) : Optimisation de transformée rapide
    structurée en arbre convolutionnel – Rémy Abergel (4pm) : The Shannon
    total variation
  • May 12, 2016 (MAP5, salle de conseil de l’aile Turing) – SMATI seminar
    – Cécile Louchet (3pm) : Statistique fonctionnelle et quelques
    applications à l’image – Guillaume Tartavel (4pm) : Wasserstein Loss for
    Image Synthesis and Restoration
  • March 2: Luca de Masi – Segmentation of pelvic pediatric MRI.
  • March 20: Cécile Muller – Pelvis neural network : MRI neurotractography vs Lightsheet microscopy
  • May 2: Hélène Urien – Brain lesion detection in 3D PET images using max-trees and a new spatial context criterion
  • May 18: Hadrien Bertrand, Alessio Virzi: ISBI overview
  • October 13, 3pm: W. Pieczynski (Telecom Sud Paris) – Modèles de Markov triplets en signal et image
  • November 24, 4:30pm: Pietro Gori – Learning the structural organization of anatomical shape complexes
  • November 29, 3pm: Gerard Sanroma – MSClique: Multiple structure discovery through the maximum weighted clique problem
  • November 29, 4pm: Ahmad Chaddad – Radiomic application on medical images
  • November 30, 9:30am: Stian Normann Anfinsen, Carlos Lopez-Martinez, Philippe Réfrégier
  • December 1, 2pm: Alba Garcia – Content-based fMRI brain maps retrieval


  • 2015
  • January 23, 2015, 11:00am – Henri Maître (F900) : De la photo argentique à la photo computationnelle.
  • January 15, 2015, 3pm – Joint TII-MAP5 seminar (C49) – Lionel Moisan : MAP, LSE, ICE : 3 variantes pour la régularisation par variation totale – Vincent Duval : Garanties théoriques pour les méthodes de déconvolution avec a priori de parcimonie.
  • February 5, 2015, 3pm – Joint TII-MAP5 seminar (Paris Descartes, salle du conseil, 7e étage) – Agnès Desolneux : quand l’approche a
    contrario devient générative – Marc Lebrun : From Theory to Practice: a Tour of Image Denoising
  • Jeudi 21 avril à 10h30 en salle C49 – Rémy Abergel – Méthodes duales pour la minimisation de la variation totale
  • Jeudi 31 mars à 10h15 en salle C49 – Steve Oudot – Topological Descriptors for Geometric Data Comparison
  • Jeudi 11 février 2016 à 10h15 (en C49) – Isabelle Bloch – Mathematical morphology from an algebraic point of view
  • Jeudi 19 novembre à 10h30 (en amphi Estaunié) – Frédéric Dufaux – HDR Video Coding: challenges and recent research activities
  • Jeudi 22 octobre à 10h30 (en C49) – Gersende Fort – Méthodes de Monte Carlo par Chaînes de Markov
  • 25 juin – Bruno Vallet (IGN) – Collecte de données 3D pour la modélisation urbaine à l’IGN, résultats, verrous et tendances
  • 16 avril – Isabelle Bloch – Probabilités et modélisation de l’incertitude : un point de vue historique
  • 1 avril 2015 – Laurent Mugnier (ONERA) – Optique adaptative et applications
  • March 17, 2015, 2pm (C06) – Carlos Lopez-Martinez (Barcelona) : PolSAR and PolInSAR Temporal Series Analysis with Binary Partition Trees – Hong Sun (Wuhan)
  • September 13, 10am: Presentation of papers from Medical Imaging Analysis, vol. 33
  • September 20, 3pm: Presentation of papers from Medical Imaging Analysis, vol. 33
  • September 27, 10am: Presentation of papers from Medical Imaging Analysis, vol. 33
  • October 11, 3pm: Hélène Urien – MICCAI MSSEG Challenge
  • October 18, 10am: Alessio Virzi – Pelvic bone segmentation in pediatric MRI
  • November 21, 4pm: Thais Roque (univ. Oxford) – Medical Imaging
    Based Modelling of Tumour Growth: Towards Personalised Predictions – Ruben Sanchez – Presentation of papers from JFR 2016
  • November-December: See general team seminars by Pietro Gori, Gerard Sanroma, Ahmad Chaddad, Alba Garcia.


Séminaire Parisien des Mathématiques Appliquées à  l’Imagerie.

Computer graphics

Medical imaging, digital health and spatial reasoning:

  • 2021
  • January 11: Robin Kips – Learning to transform the color of objects: color controllable Generative Adversarial Networks.
  • January 25
  • February 2: Adrien Parrot, Interhop
  • February 22: Francesco Maso: Anatomically constrained Cross-domain CT image translation using CycleGAN
  • March 1st: Jules Françoise and Baptiste Caramiaux: Marcelle, a new interactive machine learning toolkit.
  • March 15: Fabienne Orsi: Communs et santé, un impossible amour ?
  • May 5: Feedback on ISBI 2021 – Benoît Dufumier : Contrastive Learning with Continuous Proxy Meta-Data for 3D MRI Classification
  • May 17: Giammarco La Barbera: Geometric and Appearance Domain Adaptation for Pediatric Image Segmentation
  • June 28: Robin Louiset: UCSL : A Machine Learning Expectation-Maximization framework for Unsupervised Clustering driven by Supervised Learning
  • August 30: Feedback from summer schools, general discussion
  • 2020
  • February 3: Papers from MICCAL Conference
  • February 24: round table discussion
  • April2: Corentin Mercier – Fast APSS
  • April 24: round table discussion
  • May 15: Yanis Djebra – MR based PET motion correction
  • May 29: Yanis Chemli – Motion Correction for brain PET using a Real Time Motion Capture System — Karine Haddadi – Simulation de données de patients pour la comparaison des performances cliniques de systèmes d’absorptiométrie biphotonique à rayons X
  • June 20: round table discussion
  • June 25: round table discussion
  • September 7: Matthis Maillard – Knowledge distillation from multi-modal to mono-modal segmentation networks — Francesco and Ilaria,  new interns — Bruno Belucci, final internship presentation
  • September 21: Mateus Riva – Feedback from ECML
  • October 19: Rebeca Vetil –  Pancreas Segmentation : some experiments on the 2D UNet — Feedback from MICCAI
  • November 2:  Camille Ruppli
  • November 16: Antoine Pirovano – Improving Interpretability for Computer-aided Diagnosis tools on Whole Slide Imaging with Multiple Instance Learning and Gradient-based Explanations
  • December 3: Isabelle Bloch – Applications of Mathematical Morphology to Symbolic Music Representations — Mateus Riva – Approximation of dilation-based spatial relations to add structural constraints in neural networks
  • December 14: Corentin Mercier – Rehearsal of PhD defense
  • 2019
  • February 28, 2:30pm: Pavlo Mozharovskyi – Depth for curve data and applications
  • May 6: Papers from ISBI Conference
  • 2018
  • January 15, 9:30am: Alessandro Delmonte, Corentin Mercier – Summary of CoBCoM Winter School Workshop. Alessio Virzi, Cécile Muller – Summary of Surgetica 2017
  • January 24, 9am: Hélène Urien – Détection et segmentation de lésions dans des images cérébrales TEP-IRM
  • Mars 13, 2pm: Cécile Muller – Development of pelvic neurotractography
    for image guided surgery in children with tumors and malformations
  • March 29, 9:30am: Ruben Sanchez de la Rosa – Reconstruction, improvement and analysis of angiotomosynthesis images to optimize their clinical performances
  • September 5, 2pm: Maxime Fondin – Segmentation et comptage de follicules ovariens en IRM
  • November 22, 2pm: Pietro Gori, Hadrien Bertrand, Corentin Mercier: Summary of MICCAI and VBCM conferences
  • December 12, 10am: Hadrien Bertrand: Hyper-parameter optimization in deep learning and transfer learning – Applications to medical imaging
  • 2017
  • September 14, 3pm: Hadrien Bertrand – Deep learning and transfer learning in medical imaging.
  • November 6, 3pm: Raphael Berdugo – Covariation and Non Covariation methods applied to Shape Analysis and Medical Imaging
  • November 13, 10pm: Timothée Evain – Nouvelles méthodes de segmentation en imagerie tomographique volumique à  faisceau conique dentaire


  • PhD seminar:
      1. presentation of the thesis subject at the beginning
      2. at the end of the first year
      3. mid-term evaluation
  • General team meeting: about once a weak
  • Specific working groups (ex: compressed sensing in 2009, patch-based methods in 2014, SMATI and DIFF-L in 2015-2016…)
  • Older seminars


Postal address : LTCI, Télécom Paris, IDS / IMAGES, 19 place Marguerite Perey

Copyright (c) TII – 2009, IMAGES 2019

Latest news

  • Pas d’article actuellement.No news article at the moment.