Nicolas Papadakis

Email :	nicolas.papadakis a math.u-bordeaux.fr	Institut de Mathématiques de Bordeaux
Phone :	+33 5 40 00 21 16	351 Cours de la Libération
Office :	251	33405 TALENCE Cedex, France

Nicolas Papadakis - CNRS Researcher

Home Job offers Events

Papers Talks Codes

ANR GOTMI ANR PostProdLEAP H2020 NoMADS

Supervision Teaching

Bio

News

01/2021: All talks and slides of Mathematics and Image Analysis (MIA'21) are now available online.
09/2020: The conference on Mathematics and Image Analysis (MIA'21) will be virtual. Registration is now open.
06/2020: Our Minisymposium on Semi-Supervised Learning at SIAM Conference on Imaging Science goes virtual (July 6-10).
02/2020: Our multi-disciplinary project on Machine Learning for Ocean and SWOT data processing has been awarded a CNRS 80|PRIME grant
01/2020: Talk at the Workshop on AI for Ocean, Atmosphere and Climate Dynamics in Brest

Last papers

04/2021: Overrelaxed Sinkhorn-Knopp Algorithm for Regularized Optimal Transport, MDPI Algorithms Special Issue on Optimal Transport: Algorithms and Applications
04/2021: Nonlinear Power Method for Computing Eigenvectors of Proximal Operators and Neural Networks, SIMA Journal on Imaging
03/2021: Learning local regularization for variational image restoration, SSVM'21
02/2021: Multi-task deep learning for image segmentation using recursive approximation tasks, IEEE Transactions on Image Processing
01/2021: New preprint On the Existence of Optimal Transport Gradient for Learning Generative Models
10/2020: New preprint GraphXCOVID: Explainable Deep Graph Diffusion Pseudo-Labelling for Identifying COVID-19 on Chest X-rays
09/2020: Block based refitting in ℓ₁₂ sparse regularisation, Journal of Mathematical Imaging and Vision
08/2020: Learning to segment microscopy images with lazy labels has been presented at the ECCV 2020 workshop on BioImage Computing
06/2020: Étude comparative de méthodologies issues de Mask R-CNN : Application au Corpus DeepFashion2 has been presented at RFIAP 2020
05/2020: New preprint on Wasserstein Generative models for texture synthesis
03/2020: Variational Osmosis for Non-linear Image Fusion, IEEE Transactions on Image Processing
02/2020: Image denoising method for the SWOT satellite mission, Remote Sensing

Research interests

Image Assimilation in Geosciences
Improving numerical prediction with the integration of structured image information in data assimilation systems.
Keywords: Variational and sequential assimilation, observation errors.
Application: Oceanography

images/labyrinth.gif

Numerical methods for Optimal Transportation
Definition of generalized optimal transportation models for comparing and interpolating densities
Keywords: Physical-based regularity constraints, approximations for fast computation, data analysis in Wasserstein space, Neural Networks.
Applications: Computational photography, Geosciences.

images/sct.gif

Inverse problems in Imaging sciences
Introduction of new variational and discrete models for image/video processing and analysis.
Keywords: Segmentation, superpixels, denoising, motion estimation, tracking, stereovision, 3D reconstruction, color transfer, colorization, inpainting.
Applications: Satellite data, Movie post-production, Football replay, Medical imaging.

images/teaser.png

Machine Learning
Algorithms for processing non structured data represented by graph.
Keywords: Clustering, label propagation, semi-supervized learning, unsupervised learning.
Applications: Hyperspectral and Medical data