RESEARCH INTERESTS
My research interests span the areas of deep learning, computer vision, machine learning, medical image analysis and artificial intelligence.
In a nutshell, the goal of my research is to develop efficient, scalable and mathematically well-grounded algorithms that are capable of analyzing and extracting (semantic) information from various types of visual data (be it static natural images, video, medical image data etc.).
EDITORIAL BOARD MEMBER
International Journal of Computer Vision
Computer Vision and Image Understanding Journal
Computational Intelligence Journal (Wiley)
LATEST NEWS
Excited to share our work "Online Bag-of-Visual-Words Generation for Unsupervised Representation Learning". Proposes novel approach for self-supervised learning that surpasses all previous state-of-the-art methods (including contrastive-based ones), setting new state-of-the-art results on a variety of commonly used benchmarks. Source code and trained models are available at the project page.
I will be serving as area chair at CVPR 2021
Our work on "QUEST: Quantized embedding space for transferring knowledge" has been accepted to ECCV 2020: proposes simple and effective approach that achieves new state-of-the-art results on knowledge distillation
Our work on "Learning Representations by Predicting Bags of Visual Words" has been accepted to CVPR 2020: proposes new reconstruction-based approach for self-supervised learning that is on par with contrastive-based methods
I will be serving as area chair at ECCV 2020
My student Spyros Gidaris has been awarded 2 prizes for his PhD work: the Ponts Foundation Best Thesis Prize and the University Paris-Est Best Thesis prize. Congratulations, Spyros !
Our work on "Deep Tone Mapping Operator for High Dynamic Range Images" has been accepted to the IEEE Transactions on Image Processing
Our work on "Boosting Few-Shot Visual Learning with Self-Supervision"" has been accepted to ICCV 2019 (achieves state-of-the-art performance for few-shot learning - stay tuned for the source code which will be publicly available soon)
The source of our CVPR 2019 work on "Generating Classification Weights with GNN Denoising Autoencoders for Few-Shot Learning" is now available online (github page)
My student Sergey Zagoruyko has received the AFRIF 2018 Thesis Prize for his PhD work. Congratulations, Sergey !
Our work on "Exploring Weight Symmetry in Deep Neural Networks" has been accepted to the Computer Vision and Image Understanding journal. Source code is now available (github page)
I will be serving as area chair at ICCV 2019
Our work on "Generating Classification Weights with GNN Denoising Autoencoders for Few-Shot Learning" has been accepted as an oral presentation at CVPR 2019
I will be serving as area chair at BMVC 2019
Check our PAMI paper on Scattering Networks for Hybrid Representation Learning
The source code of our CVPR paper "Dynamic Few-Shot Visual Learning without Forgetting" has been released (github page)
I will be serving as area chair at BMVC 2018
Source code is now available for our ICLR paper on "Unsupervised Representation Learning by Predicting Image Rotations". This work significantly closes the gap between unsupervised and supervised learning for several core visual tasks (github page)
Source code is publicly available for our work on "Rotation Equivariant Vector Field Networks" (github page)
Our work "Deep Compare: A Study on Using Convolutional Neural Networks to Compare Image Patches" has been published in CVIU journal
I will be serving as area chair at CVPR 2018
PyTorch implementation of our paper on "Dynamic Edge-Conditioned Filters in Convolutional Neural Networks on Graphs" is now available. Check the project page
Check our ICCV 2017 paper "Rotation Equivariant Vector Field Networks"
Check our paper "DiracNets: Training Very Deep Neural Networks Without Skip-Connections". Source code is available at the project page
Our work on structured prediction using deep neural networks is available "Detect, Replace, Refine: Deep Structured Prediction For Pixel Wise Labeling"
Our work on generalizing CNNs to graph-structured problems is available "Dynamic Edge-Conditioned Filters in Convolutional Neural Networks on Graphs"
3 papers accepted at CVPR 2017
I will be serving as area chair at BMVC 2017
Source code is now available for our latest ICLR 2017 paper: "Paying More Attention to Attention: Improving the Performace of Convolutional Neural Networks via Attention Transfer"
[Project page]
I will be area chair at CVPR 2017
Source code of our state-of-the-art box proposal generation method (AttractioNet) now available
[Source code and project page]
3 papers accepted at BMVC 2016
1 paper accepted at MICCAI 2016
Our new "Wide Residual Networks" architecture outperforms in accuracy and efficiency all previous deep residual networks, showing also that deeper is not necessarily better. Significantly improves the state-of-the-art on CIFAR10, CIFAR100 and SVHN.
[Source code][Technical report]
Our paper "LocNet: Improving Localization Accuracy for Object Detection" was accepted as oral presentation at CVPR 2016. The code and extended technical report are available.
[Source code][Extended technical report]
I will be serving as an area chair for CVPR 2016
Our paper "Object Detection via a Multi-region and Semantic Segmentation-aware CNN Model" will appear at ICCV 2015. Sets new state-of-the-art for object detection:
[Source code][Extended technical report]
Our paper "HARF: Hierarchy-associated Rich Features for Salient Object Detection" was accepted at ICCV 2015.
(Pdf and code will be made available here shortly)
Code (Torch, Matlab, C++) and trained models are publicly available for our CVPR2015 paper on learning to compare patches via convolutional neural networks (which provides state-of-the-art results). Check the project page for mode details.
I will be serving as an area chair for ICCV 2015
Our new tutorial paper with Jean-Christophe Pesquet entitled Playing with Duality: An Overview of Recent Primal-Dual Approaches for Solving Large-Scale Optimization Problems will appear as a feature article in IEEE Signal Processing magazine.
Our new TPAMI journal paper describes a framework for efficient structured max-margin learning of high-order models.
New NIPS paper accepted: shows how to to provide an order of magnitude speed-up for graphical model optimization through a coarse-to-fine cascade of pruning classifiers (paper and source code will very soon be publicly available).
I joined the editorial board of the International Journal for Computer Vision (IJCV).
Since 2013, I serve in the editorial board of the Journal for Computer Vision and Image Understanding (CVIU).
The schedule for our CVPR 2014 tutorial is now available.
I am a guest editor of the IJCV special issue on "Graphical Models for Scene Understanding". See the call for papers.
Giving a tutorial about "Discrete Optimization in Computer Vision" on September 10 at the International Conference on Image Analysis and Processing (ICIAP) 2013.
The website of our ICCV 2013 workshop on "Graphical models for scene understanding: challenges and perspectives" has been launched.
Organizing on July 8 a "Networks, Optimization and Vision" colloquium at Ecole des Ponts ParisTech. Check the event's web page for registering as well as for obtaining more information about the program.
Giving an invited talk on July 4 at the EURO 2013 conference about "Efficient Learning of High Order Graphical Models for Computer Vision and Image Analysis".
The paper "A Comparative Study of Modern Inference Techniques for Discrete Energy Minimization Problems" was accepted as an oral presentation at CVPR 2013. You can also check the corresponding web-site with the new benchmark database of problems.
Our paper "MRF-based Blind Image Deconvolution" is now available. It describes a super efficient and effective algorithm that can recover large and complicated blur kernels. Check it out here.
Keynote speaker at the ECCV 2012 Workshop on Higher-Order Models and Global Constraints in Computer Vision.
Giving a tutorial on "Inference and Learning for Image Processing and Computer Vision" at the EUSIPCO 2012 international conference.
Serving as area chair of the 11th Asian Conference on Computer Vision (ACCV2012).
Invited to be the tutorial speaker of the MIUA 2012 conference (16th conference on Medical Image Understanding and Analysis) .
The slides of my tutorial "An introduction to graphical models and discrete optimization" for the Biomedical Image Analysis Summer School 2012 are now available ([Part I], [Part II]).