Yousri Kessentini 

Senior researcher at CRNS & head of DeepVision research team

Certified as an official Deep Learning instructor & ambassador from NVIDIA DLI

Email: yousri.kessentini<at>crns.rnrt.tn

           yousri.kessentini<at>gmail.com

Phone: +216 74863042         Fax: +216 74863041

Address: CRNS, Technopole of Sfax, PO Box 275,  Sakiet Ezzit, 3021 Sfax, Tunisia.

BIOGRAPHY

I am currently Associate Professor (HDR) at the Digital Research Center of Sfax (CRNS) and the head of the DeepVision research team. I am also a member of the SM@RTS Laboratory. I obtained my accreditation to supervised research (Habilitation Universitaire) in 2021 from the university of Sfax, my Ph.D. degree from the university of Rouen in 2009 and my engineer's degree from the National Engineering School of Sfax (Tunisia). Before joining CRNS, i was assistant professor at ISIMA and postdoctoral researcher at ITESOFT and LITIS

I am certified as an official instructor and ambassador from the NVIDIA Deep Learning Institute. My main research areas concern computer vision, deep learning, document processing and recognition. I have participated in several research projects and technology transfer projects. I have published more than 60 papers in international conferences and journals. I am also a reviewer for several international conferences and journals in the field of pattern recognition and computer vision.

 

NEWS

New accepted paper in WACV conference  (Rank A) 

Our paper Graph Neural Networks for End-to-End Information Extraction from Handwritten Documents has been accepted in Winter Conference on Applications of Computer Vision (WACV'2024). 

You can access it using this link:  [Link] 

New accepted paper in Neurocmputing  (Q1, IF=6) 

Our paper A Two-stream SpatioTemporal Fusion Transformer for Very High Resolution Satellites Images has been published in the Neurocomputing journal in October 2023. 

You can access it using this link:  [Link] 

New accepted paper in the AAAI'2023 conference  (Rank A*) 

Our paper Text-DIAE: A Self-Supervised Degradation Invariant Autoencoder for Text Recognition and Document Enhancement has been accepted in the Thirty-Seventh AAAI Conference on Artificial Intelligence (AAAI'2023).

AAAI is one of the leading international academic conference in AI ranked A*.

This year 8,777 papers have been submitted. The overall acceptance rate is 19.6%.

Link to our paper: https://arxiv.org/abs/2203.04814  

New accepted paper in Pattern Recognition Letters Journal (Q1, IF=5.1) 

Our paper Few Shots Are All You Need: A Progressive Few Shot Learning Approach for Low Resource Handwriting Recognition, has been published in the journal Pattern Recognition Letters (PRL) in June 2022. 

You can access it using these links:  [Link] [code] [arxiv] 

New accepted paper in ESWA journal (Q1, IF=8.665)

Our paper "Masking for better discovery: Weakly supervised complementary body regions mining for person re-identification" has been published in the Expert Systems with Applications journal in February 2022.

You can access to the paper using this link : [Link]  [PDF] 

New accepted paper in Pattern Recognition Letters Journal (Q1, IF=5.1) 

Our paper Transformer-Based Approach for Joint Handwriting and Named Entity Recognition in Historical documents, has been published in the journal Pattern Recognition Letters (PRL) in November 2021. 

You can download it for free using this link: [Link] [PDF] 

The version 2 of DE-GAN is now published in the Pattern Recognition Journal (Q1, IF=8.518) 

In this paper "Enhance to read better: A Multi-Task Adversarial Network for Handwritten Document Image Enhancement", we present an improved Generative Adversarial Network for Handwritten Document Image Enhancement based on multi-task learning. The proposed architecture integrates a handwritten text recognizer that promotes the generated document image to be more readable.  [Link] [arxiv] [DOI]

New accepted paper in ESWA journal (Q1, IF=8.665) 

our paper "Pansharpening approach via two-stream detail injection based on relativistic generative adversarial networks" has been published in the journal Expert Systems with Applications (Q1, IF=6.954).

You can download it for free using this link: https://authors.elsevier.com/c/1dxi73PiGTI070 

The link is valid until December 09, 2021 

New accepted paper in NCAA journal IF 5.606 

Our paper Domain and writer adaptation of offline Arabic handwriting recognition using deep neural networks is accepted for publication in Neural Computing and Applications (NCAA), September 2021. Impact Factor: 5.606  

DOI: https://doi.org/10.1007/s00521-021-06520-7 

New accepted paper in IEEE TPAMI journal IF=24.314

Our paper "DE-GAN: A Conditional Generative Adversarial Network for Document Enhancement" is accepted in IEEE Transactions on Pattern Analysis and Machine Intelligence (T-PAMI), March 2022. Impact Factor=24.314

[Link] [code] [arxiv] 

I defended my habilitation to supervised research (Habilitation Universitaire)

I obtained my habilitation degree (Habilitation Universitaire) on 21th May 2021 from the university of Sfax.

Title : Contributions au traitement automatique des images de documents et de scènes à l'ère de l'apprentissage profond.


The MedPRAI proceeding book is online 

This book constitutes the refereed proceedings of the 4th Mediterranean Conference on Pattern Recognition and Artificial Intelligence, MedPRAI 2020, held in Hammamet, Tunisia, in December 2020.​ Due to the COVID-19 pandemic the conference was held online.The 24 revised papers presented were thoroughly reviewed and selected from 72 submissions. The papers are covering the topics of recent advancements in different areas of pattern recognition and artificial intelligence.

DOI : https://doi.org/10.1007/978-3-030-71804-6

New accepted paper in Applied Soft Computing

Our paper "Federated learning for COVID-19 screening from Chest X-ray images" is accepted in Applied Soft Computing journal, Mars 2021. Impact Factor=8.263, Q1

In this work, we present a collaborative federated learning framework allowing multiple medical institutions screening COVID-19 from Chest X-ray images using deep learning without sharing patient data. We investigate several key properties and specificities of federated learning setting including the not independent and identically distributed (non-IID) and unbalanced data distributions that naturally arise.

DOI: https://doi.org/10.1016/j.asoc.2021.107330 

Our grant proposal is accepted – STINT Sweden

Our Project “DocPRESERV: Preserving & Processing Historical Document Images with Artificial Intelligence” is accepted in the Initiation Grants for Internationalisation program financed by STINT (The Swedish Foundation for International Cooperation in Research and Higher Education). 

Partner: Blekinge Institute of Technology in Karlskrona (Sweden)

Best Student Paper Award at ICPR'2020 

Our paper “A Few-shot Learning Approach for Historical Encoded Manuscript Recognition" won the Best Student Paper Award at the 25th International Conference on Pattern Recognition.The Overall Oral acceptance rate in ICPR’20220 is 4,4% (144 over 3250 submissions)

 Link to our paper: https://arxiv.org/abs/2009.12577

Other news...