Research Projects
Title: Parkinson Diagnostics Collaboration Platform
Project Manager: Adel M. Alimi
Développement d'une plateforme collaborative pour le suivi et le diagnostic de la maladie de Parkinson en utilisant la théorie Bêta Elliptique
Title: Intelligent prediction of ADHD neurodevelopmental disorder in children by multimodal analysis
Project Manager: Rim Walha
Attention Deficit Hyperactivity Disorder (ADHD) is currently one of the most common neurodevelopmental disorders. The overall objective of this project is to harness artificial intelligence for the benefit of public health, specifically focusing on the early prediction of ADHD in children. More precisely, this project aims to implement an artificial intelligence-based solution to predict ADHD in children at an early stage. This solution will incorporate multimodal analysis using various sources of information, such as responses to a questionnaire, EEG signals, handwriting analysis, etc.
Title: Real-time end-to-end text detection and recognition in the wild
Project Manager: Adel M. Alimi
The project focuses on the development of a real-time end-to-end text detection and recognition system designed for diverse and dynamic real-world scenarios, often referred to as "in the wild." The primary goal is to create a robust and efficient solution that can accurately detect and recognize text in images captured under challenging conditions, such as varying lighting, blur, low resolution, and diverse backgrounds. The system comprises two key modules: the first is dedicated to accurately localizing text in images, ensuring resilience to common degradations, while the second is designed to efficiently recognize words across different fonts and scripts. The challenges addressed include the need for robustness in the face of adverse conditions and the demand for real-time processing capabilities. Additionally, the proposed system aims to be lightweight and adaptable for deployment on mobile devices, marking a significant advancement in extending text detection and recognition capabilities beyond controlled environments to the complexities encountered in real-world settings. The project, titled "Real-time end-to-end Text Detection and Recognition in the Wild," is led by a collaborative team with expertise from Tunisia and India. The Tunisian team is spearheaded by Adel Alimi, serving as the principal investigator, alongside Fadoua Drira, who contributes as a co-investigator, and Rim Walha, a valuable team member. Complementing this, the Indian team is led by Umpada Pal as the principal investigator, with Ujjwal Bhattacharya serving as the co-investigator. Together, these interdisciplinary teams aim to address the complex challenges of developing a real-time end-to-end text detection and recognition system capable of operating seamlessly in diverse and dynamic real-world environments. Main publications: [1] Riadh Harizi, Rim Walha, Fadoua Drira, Mourad Zaied: Convolutional neural network with joint stepwise character/word modeling based system for scene text recognition. Multim. Tools Appl. 81(3): 3091-3106 (2022) [2] Riadh Harizi, Rim Walha, Fadoua Drira: Deep-learning based end-to-end system for text reading in the wild. Multim. Tools Appl. 81(17): 24691-24719 (2022)