- Statut : Ingénieur
- louisa.kessi univ-lr.fr
Doctor of science in Computer Science and Applied mathematics with an emphasis of machine learning and computer vision at LIRIS labs of INSA-Lyon, France.
Machine learning and applied mathematics scientist at L3I labs of university of La Rochelle.
Machine learning scientist with strong quantitative background and experience in statistical machine learning and computer vision with an emphasis on unsupervised Learning, developing tasks that drive visual learning without requiring manual annotation, Discovering the objects, object parts, and other patterns from visual data without human intervention, Pre-processing of color images, Color segmentation, Color feature extraction, Modelisation of spatial relations, Distances and metrics in high dimensional spaces, Incremental clustering of large database, Part-based recognition , Logical structures recognition for Heterogeneous document, Sampling and estimation, intrusion and Fraud detection.
About me :
I am a Machine Learning and Applied Mathematics scientist under the European project FEDER SeAD. at L3i, University of La Rochelle.
I work with : Mickael Coustaty and Gomez-Krämer Petra.
Research team : Images et contenus
The aim of the project is the development of new methods and algorithms in the domains of steganography and digitization/numerisation of secured diploma’s copies.
I have achieved a PhD titled "Unsupervised Detection based on Spatial Relationships, Application for Object Detection and Recognition of Colored Business Document Structures" in the LIRIS CNRS Laboratory at the National Institute of Applied Sciences (INSA de Lyon) within the computer vision team, under the supervision of Christophe Garcia and Frank Le Bourgeois. I have worked on an industrial project named "Document On Demand" for the company ITESOFT, one of the main European leader in the digitization of business documents, renamed YOOZ for US customers.
Before coming to INSA-Lyon, I have received a Master Research degree in Electrical Engineering, Signal processing , Networking and Telecommunication with excellent distinction and a License degree in Electrical Engineering and Applied mathematics with excellent distinction at Mouloud MAMMERI University, Algeria. I also spent several months as Research scientist at Centre for Development of Advanced Technologies (C.D.T.A) to participate on the research project’’ Multimodal Biometric System (voice, face an Fingerprint)’’ ) , Research collaboration between C.D.T.A and Algerian National Gendarmerie in the framework of my Master research Thesis .
Professional Experience :
***4 years and half at LIRIS and ITESOFT-yooz (document digitizing company), research and development , under the proposed thesis project. Lyon/Aimargues
thesis defense : 13 september 2018
Association : LIRIS Labs,Laboratory of InfoRmatics in Image and Information Systems
Pole : Computer Vision Pattern Recognition
Team imagine : Feature Extraction and Identification
National Institute of Applied Sciences of Lyon
Thesis Title : Unsupervised Detection based on Spatial Relationships – Application for Object Detection and Recognition of Colored Business Document Structures
Thesis Advisor :
Christophe Garcia (Prof. of Computer Science), Deputy Director of the LIRIS Laboratory, Vice-President for Research in Information and Digital Society of INSA Lyon.
Frank Le Bourgeois (Prof. of applied mathematics & Computer Science).
Thesis developments and contributions :
Development of the first generic color segmentation engine called AColDPS for noisy and degraded color documents worked in high dimensional space.
_Introduction of a new method called Non Local Means Inter-Image (NLM-I) to separate the preprinted form from the added text. This is the first alignment method that is accurate at the pixel level which allows spatial distortion of the paper support.
_Development of new image restoration of broken characters after the separation between texts and graphics when they are overlapped system .
Development of the first generic and scalable unsupervised engine for the recognition of heterogeneous logical structures for digitized business documents in high dimensional space. Modelization of spatial relationships ,Introduce part-based recognition principle on high dimensional space , high dimensional feature space, introduce of new distances and metric working on high dimensional space and
Introduce new unsupervised approaches for both the modelization of logical structures of color business documents and Object Recognition (PASCAL /IMAGENET) .
Worked on large database in high dimensional feature space for both business document structure and object recognition applications.
During my doctoral Thesis , I have the chance to :
*** In Jul’15–oct’15 Participate as ML Scientist in the project GUWENSHIBIE, French ANR research project, research collaboration between LIRIS labs of INSA of Lyon and Tsinghua university.
Adaptation and generalization of the developed engine " AColDPS" initially developed for business documents on historical noisy Chinese document.
***Oct’15–oct’15 Visiting Researcher. Peking university of China , Tsinghua university.
Invited as a speaker to expose my research work on the field of color information analysis and how to generalize and adapt the developed software" AColDPS" on historical noisy Chinese document for the research project GUWENSHIBIE ( French ANR research project, research collaboration LIRIS labs of INSA of Lyon and Tsinghua university.
*** Aug’15–sep’15 Visiting Researcher. University of Kyoto.
Selected to participate and follow the courses of the prestigious Machine Learning Summer Schools at Kyoto, Japan and expose my research work.
Before my doctoral thesis I was :
*** In Apr’12–feb’13 Data Science Researcher Intern , Centre for Development of Advanced Technologies.Algiers. Algeria
Research Project’’ Multimodal Biometric System (voice, face and Fingerprint)’’
Research Team ’’Biometric and Security Multimedia’’
Division "Systems Architecture and Multimedia’’
*** In Apr’10–oct’10 Data Science Research Intern on Telecom and Network workplace. ALGERIE TELECOM group.