Avatar

Frédéric RAYAR, PhD

                 

Post-doctoral Researcher
ELIPSE Team
Toulouse Institute of Computer Science Research (IRIT)
Toulouse, FRANCE
frederic {dot} rayar {at} irit {dot} fr



About me

My research focus on the combination of Artificial Intelligence and Machine Learning with Human-Computer Interactions and Information Visualisation. Applications of my contributions occur in fields such as Digital Humanities, Assistive Technologies and Education.

Since October 2018, I am a post-doctoral researcher in the University of Toulouse, where I combine Artificial Intelligence and Human-Computer Interactions to help Visually Impaired People life improvment (especially, the access to culture and education).

Previously, I was in Kyushu University (Fukuoka, Japan) for one year (2017-2018), collaborating with Pr. Seiichi Uchida on several topics related to Deep Learning (its understanding, explanation and usage).

From 2011 to 2017, I was in the Computer Science Laboratory of Tours (France), where I was successively a R&D engineer and a PhD student, while being a part-time lecturer.

Before my research activities, I worked for one year in Synapse Développement (Toulouse, France), as a software developer, right after receiving my M.S. in Computer Science in 2009 from the University of Toulouse.


Publications

2019

  • Frédéric Rayar and Seiichi Uchida. A visual matrix image-based representation for fast graph classification. Pattern Recognition : under review.
  • Frédéric Rayar and Seiichi Uchida. Comic text detection using neural network approach. International Workshop on coMics ANalysis, Processing and Understanding, MANPU@MMM 2019 : 672-683

2018

  • Frédéric Rayar and Seiichi Uchida. An Image-Based Representation for Graph Classification. International Workshops on Statistical Techniques in Pattern Recognition (SPR) and Structural and Syntactic Pattern Recognition (SSPR) 2018 : 140-149.
  • Frédéric Rayar and Seiichi Uchida. On Fast Sample Preselection for Speeding up Convolutional Neural Network Training. International Workshops on Statistical Techniques in Pattern Recognition (SPR) and Structural and Syntactic Pattern Recognition (SSPR) 2018 : 65-75.
  • Frédéric Rayar, Masanori Goto and Seiichi Uchida. CNN training with graph-based sample preselection: application to handwritten character recognition. International Workshop on Document Analysis Systems (DAS) 2018 : 19-24.
  • Frédéric Rayar, Sabine Barrat, Fatma Bouali and Gilles Venturini. A Viewable Indexing Structure for the Interactive Exploration of Dynamic and Large Image Collections. ACM TKDD Special Issue on Interactive Data Exploration and Analytics (IDEA) 2018 : 1-26.

2017

  • Frédéric Rayar. Accessible Comics for Visually Impaired People: Challenges and Opportunities. International Workshop on coMics ANalysis, Processing and Understanding, MANPU@ICDAR 2017 : 9-14.
  • Frédéric Rayar. Interface tangible pour la manipulation d’objets 3D dans des expositions publiques. A la Conférence francophone sur l’Interaction Homme-Machine, IHM 2017 : 199-206.
  • Frédéric Rayar and Pierre Robillard. ORBox: conception d’une surface intelligente embarquée pour la reconnaissance de petits objets. A la Conférence francophone sur l’Interaction Homme-Machine, IHM 2017 : 179-186.
  • Frédéric Rayar and Zeina Abu-Aisheh. Photo(Graph) Gallery: An "exhibition" of graph classification. International Conference on Information Visualisation (IV) 2017 : 262-267.

2016

  • Frédéric Rayar. Interactive, incremental and multi-level exploration of large collections of images. PhD Thesis.
  • Frédéric Rayar, Sabine Barrat, Fatma Bouali and Gilles Venturini. APoD eXplorer: Recommendation System and Interactive Exploration of a Dynamic Image Collection. International Conference on Information Visualisation (IV) 2016 : 118-123.
  • Frédéric Rayar, Sabine Barrat, Fatma Bouali and Gilles Venturini. Incremental hierarchical indexing and visualisation of large image collections. European Symposium on Artificial Neural Networks, Computational Intelligence and Machine Learning (ESANN) 2016: 659-664.
  • Frédéric Rayar, Tanmoy Mondal, Sabine Barrat, Fatma Bouali and Gilles Venturini. Visual Analysis System for Features and Distances Qualitative Assessment: Application to Word Image Matching. International Workshop on Document Analysis Systems (DAS) 2016: 381-386.
  • Frédéric Rayar, Sabine Barrat, Fatma Bouali and Gilles Venturini. Construction incrémentale d’une structure hiérarchique pour l’exploration visuelle et interactive de larges collections d’images. Conférence internationale sur l’extraction et la gestion des connaissances (EGC) 2016: 327-332.

2015

  • Frédéric Rayar, David Boas and Rémi Patrizio. ART-Chess: A Tangible Augmented Reality Chess on Tabletop. In the 10th ACM International Conference on Interactive Tabletops and Surfaces (ITS) 2015: 229-233.
  • Frédéric Rayar and Armand Renaudeau. ART-Chess: A Tangible Augmented Reality Chess on Tabletop. In the 10th ACM International Conference on Interactive Tabletops and Surfaces (ITS) 2015: 373-378.
  • Frédéric Rayar, Sabine Barrat, Fatma Bouali and Gilles Venturini. An Approximate Proximity Graph Incremental Construction for Large Image Collections Indexing. International Symposium on Methodologies for Intelligent Systems (ISMIS) 2015: 59-68.
  • Partha Pratim Roy, Frédéric Rayar and Jean-Yves Ramel. Word Spotting in Historical Documents using Primitive based Dynamic Programming. Image and Vision Computing 2015: 15-28.

2014

  • Frédéric Rayar, Pascal Bourquin, Jean-Yves Ramel, Rémi Jimenes, Toshinori Uetani, Sandrine Breuil and Marie-Luce Demonet. PaRADIIT Project: Main Concepts and Outcomes. In the Eleventh IAPR International Workshop on Document Analysis Systems (DAS) 2014.
  • Octavio Razafindramanana, Frédéric Rayar and Gilles Venturini. Incremental Delaunay triangulation construction for clustering. In the Twenty second International Conference on Pattern Recognition (ICPR) 2014: 1354-1359.

2013

  • Jean-Yves Ramel, Nicolas Sidère and Frédéric Rayar. Interactive layout analysis, content extraction and transcription of historical printed books using Pattern Redundancy Analysis. Literary and Linguistic Computing, 2013: 301-314.
  • Octavio Razafindramanana, Frédéric Rayar and Gilles Venturini. Alpha-Approximated Delaunay Triangulation Based Descriptors for Handwritten Character Recognition. In the Twelfth Internationale Conference on Document Analysis and Recognition (ICDAR) 2013: 440-444.

2012

  • Partha Pratim Roy, Frédéric Rayar and Jean-Yves Ramel. An efficient coarse-to-fine indexing technique for fast text retrieval in historical documents. In the Tenth IAPR International Workshop on Document Analysis Systems (DAS) 2012: 150-154.
  • Frédéric Rayar, Partha Pratim Roy and Jean-Yves Ramel. Approche multi-niveaux pour la recherche de texte par l'exemple dans les images de documents historiques. Au Colloque International Francophone sur l’Ecrit et le Document (CIFED) 2012.

Administration

Research papers reviewing

  • Reviewer for International Conferences:
    • ICDAR 2017, ICDAR 2015, ICDAR 2013
    • ICPR 2018, S+SPR2018
    • DAS 2018, DAS 2016
  • Program comitee for SDNRI 2016 - RJC and SDNRI 2014 - RJC.

Events organization

  • LI-LIFO PhD Student Day 2015 (27/11/2015)
    Website
  • LI-LIFO PhD Student Day 2014 (05/11/2014)
    Website

Teaching

2016-2017

  • Graph and its applications (M2 level)
    Lectures (soon...)
  • Conception and management of a Website (M2 level)
  • Data analysis (M1 level)
  • Algorithmic Language (L3 level)
  • Introduction to LaTeX (L1 level)

  • Supervision of several Final Year Projects (M2 level)
    • Custom object recognition and tracking for Augmented reality applications. (Jessé Banah)
    • Conception and development of a custom object recognition box. (Pierre Robillard)
    • Conception and development of a custom ticketing platform. (Dylan Lecomte)

2015-2016

  • Introduction to Software Engineering (L3 level)
    Versionning system (SVN), Documentation (Doxugen), etc.
    Lecture (soon...)
  • Introduction to LaTeX (L1 level)

  • Supervision of several Final Year Projects (M2 level)
    • Web-based interactive visualization of structured information.
    • Development of the games for the reeducation of the wrist with the Sphero.
  • Supervision of several Development Project (L3 level)
    • Development of a Dobble cards generator.
    • Development of the Quarto game on the Microsoft PixelSense.
    • Development of the Tangram game on the Microsoft PixelSense.

2014-2015

  • Pattern Recognition (M2 level)
    Decision tree - Structural methods - Fuzzy logic - Feature selection and extraction - SVM
    TP LibSVM - TP Weka

  • Supervision of several Final Year Projects (M2 level)
    • Interactive visualization of large graphs
    • Objects Recognition of objects put on the Microsoft PixelSense.

2013-2014

  • Supervision of several Final Year Projects (M2 level)
    • Images Database management system (DBMS) and Incremental neighborhood graphs.
    • Merging of two clustering and transcription software.
    • Objects Recognition of objects put on the Microsoft PixelSense.
  • Supervision of a Robotic Project (M1 level)
    Virtual labyrinth for the the Orbotix Sphero

2012-2013

  • Algorithmic Language (L3 level)
    TD5-6 - TD7-8 - TD18 - TD19 - TD20 - TD21

  • Supervision of a Final Year Project (M2 level)
    Historical Document Images Exploration and Annotation with the Microsoft PixelSense
  • Supervision of a Development Project (L3 level)
    Exploiting the TagVisualizer of the Microsoft PixelSense
  • Supervision of an Operating System Project (L3 level)
    Exploiting the Orbotix Sphero to manupulate a 3D object
  • Supervision of an Operating System Project (L3 level)
    Exploiting the Orbotix Sphero to replace the mouse
  • Supervision of an Electronic Project (M1 level)
    Development of 2 Theremin (one full electronic and antoher with an Arduino)

2011-2012

Software

RETRO 2012 v2.5.1

RETRO 2012 v2.5.1

The PaRADIIT (Pattern Redundancy Analysis for Document Image Indexing and Transcription) research project focused on layout analysis, text/graphics separation, Optical Character Recognition (OCR) and text transcription processes dedicated to old and precious books.

The originality of the work relies upon the analysis and exploitation of pattern redundancy in documents to enable the efficient indexing and quick transcription of books and the identification of typographic materials.

You can obtain more information about the PaRADIIT project

RETRO 2012 is one of the software of the PaRADIIT project.
Initially dedicated to transcription, it embed now tools regarding:

  • Visualization, for segmentation validation
  • Clustering, works with characters and images
  • Transcription, assisted or automatic transcription
  • Typography studies, such as font creation body height measurement
Developed in C#, it requires Microsoft .NET Framework 3.5, and therefore aims for Windows OS (XP and above).
The software is provided under LGPL v.3 license.
You can get the code, executable, user guide, developer documentation and test data here.


An online Historical Book Explorer

Online Historical Book Explorer

The PaRADIIT (Pattern Redundancy Analysis for Document Image Indexing and Transcription) research project focused on layout analysis, text/graphics separation, Optical Character Recognition (OCR) and text transcription processes dedicated to old and precious books.

The originality of the work relies upon the analysis and exploitation of pattern redundancy in documents to enable the efficient indexing and quick transcription of books and the identification of typographic materials.

You can obtain more information about the PaRADIIT project

This Online Book Explorer, build as proof of concept, allows the users to easily interact with the digitized versions of historical books. Using a web interface, the users can navigate, visualize, search and get annotated copies of some specific elements of content extracted from historical books. They can also exploit and even enrich the available meta-data obtained with the help of PaRADIIT indexing tools by adding their own annotations on all the parts of the books.

Please, consult the Help, to quickly understand how to use the Book Explorer.


SFMitsuControl SDK 2.0

SFMitsuControl SDK 2.0

Spring 2012, Polytech'Tours purchased a Microsoft PixelSense (aka Microsoft Surface 2.0).
To exploit this brand new interactive surface computing platform, and hightlight the PaRADIIT work, I have built as a proof of concept, the application showed in the picture above.

In 2007-2008, Mitsuru Furada has developed a WPF & Silverlight FlippingBook Control (cf. his blog).
In 2009, Simon Bitschnau has ported this aplication in the Microsoft Surface, regarding the SDK 1.0 (cf. his blog).

To be able to use this control in the Samsung SUR40, second version of the table, I have adapted the application to be compatible with the SDK 2.0.

Sébastien Guillon, an engineer student in his Final Year, is currently working under my supervision, to enhance the application towards exploration and annotations of Historical Documents.

We will release an archive with FlippingBook Control for Surface SDK 2.0 (SFMitsuControls_SDK2.0), under LGPL v.3 license.

- SFMitsuControls SDK 2.0 has been released on 26th April 2013! -
Get an archive with FlippingBook Control for Surface SDK 2.0 here!


Juggling Lab for Android

Juggling Lab for Android

Juggling Lab is an desktop application for creating and animating juggling patterns. Its main goals are to help people learn juggling patterns, and to assist in inventing new ones.
Developped in java by Jack Boyce, this software is released under the GNU General Public License.
Current release: 0.6.1 (October 11, 2011).

You can obtain more information about Juggling Lab.

Developed in collaboration with Romain Richard, Juggling Lab for Android is the Android porting of Juggling Lab.
Following the main goals of the desktop version, it aims to embed a complete My Profile module, combined with social network sharing functionalities.

- v1.0 has been released on 1rst April 2013! -
Get the application for free on the Play Store!
The software is provided under LGPL v.3 license.
You can get the source and download an alpha version here.


SMSBackupRestore for Firefox OS

SMSBackupRestore for Firefox OS

SMSBackupRestore is HTML5/JavaScript application designed for Firefox OS.
It allow to save the SMS in a XML file.
Tested on a Geeksphone Keon smartphone under Boot2Gecko 1.1.0.0-prerelease.

- v1.0 has been released on September 27, 2013! -
You can get the source here.


Contact


Building IRIT 2
Office 270


IRIT, ELIPSE Team
Université Paul Sabatier,
118 Route de Narbonne,
31062 TOULOUSE CEDEX 9
Toulouse, FRANCE


Tel : +33 (0)5 61 55 67 65