Fête de la Science 2025 in Nice

Explore the 3IA Côte d'Azur program for the Fête de la Science 2025 in Nice!

  • Date: October 11, 2025
  • Times: 10:00 am to 7:00 pm
  • Place: Jardin Albert 1er (Nice)
10:00 am to 06:00 pm
  • Demo by Marco Corneli (Junior Professor in AI for History and Archaeology associated with the 3IA) on CEPAM booth
10:00 am to 1:00 pm
  • Demo: "Multi-stage CNN for fast registration of 3D preoperative CTs to 2D intraoperative X-rays" by Federica Facente (3IA Ph.D. student, Chair of Pierre Berthet-Rayne)

Abstract: Minimally invasive interventions often rely on live 2D X-rays for image guidance. Yet, anatomical localization and procedural accuracy can be enhanced by spatial alignment of these intraoperative X-rays with 3D preoperative computed tomographies (CTs). This 3D/2D registration problem is typically formulated as pose estimation of the X-ray source relatively to the CT, which is done by simulating synthetic X-rays from 2D projections of CT volumes. However, the optimization-based refinement used by the state-of-the-art deep learning approach takes several seconds, thus exceeding the allowed time budget in live image guidance. We propose LXPose (Live X-ray Pose estimation), a self-supervised multi-stage 3D/2D registration framework for real-time image guidance. LXpose removes the dependency on optimization and leverages a two-stage CNN trained with a projection loss to ensure high accuracy and computational efficiency. Moreover, we apply extensive data augmentation to mitigate the domain gap between simulated and real X-rays. Overall, LXPose yields comparable 2D registration error to the state-of-the-art method, while reducing inference time to 20ms, which demonstrates the potential of LXPose for real-time clinical deployment.
   
  • Poster:  "L’IA frugale pour sauver des vies : vers le défibrillateur de poche" ("Frugal AI to Save Lives: Towards the Pocket Defibrillator") by Rafael Silva (3IA Ph.D. student, Chair of Maxime Sermesant)

Chaque année, des millions de personnes dans le monde sont victimes d'un arrêt cardiaque soudain. Nous avons développé un système d'intelligence artificielle « frugal », capable de fonctionner sur des défibrillateurs très petits et à faible consommation, conçus pour un usage personnel ou domestique. En combinant efficacité énergétique et précision dans la détection des rythmes cardiaques dangereux, cette recherche ouvre la voie à des défibrillateurs plus accessibles, plus proches des personnes qui en ont le plus besoin.

Every year, millions of people worldwide suffer from sudden cardiac arrest. We have developed a "frugal" artificial intelligence system capable of running on very small, low-power defibrillators designed for personal or home use. By combining energy efficiency and accuracy in detecting dangerous heart rhythms, this research paves the way for more accessible defibrillators, closer to the people who need them most.
1:00 pm to 3:00 pm
2:00 pm to 4:00 pm

PEACE - Providing Explanations and Analysis for Combating Hate Expressions

Ce démonstrateur décrit une méthode consistant à détecter les discours haineux explicites et implicites dans du texte, mais également à expliquer ces prédictions avec du langage naturel (en utilisant des modèles de langages comme GPT). 

This demonstrator describes a method for detecting both explicit and implicit hate speech in text, while also explaining these predictions using natural language (by leveraging language models such as GPT).
3:00 pm to 5:00 pm
  • Demo by Edoardo Sarti (Researcher at Inria in ABS team, led by Frédéric Cazals - 3IA Chairholder)

Les protéines, molécules omniprésentes dans le vivant, sont composées par une longue séquence de petites unités nommées acides aminés. Ces séquences suivent des règles cachées, comme une grammaire biologique reliant les acides aminés à la forme finale de la molécule. Les modèles de langage, des complexes algorithmes d'Intelligence Artificielle d’abord créés pour comprendre le texte humain, ont été entraînés sur des millions de séquences de protéines pour en décrypter les secrets. Grâce à ces modèles, l’IA peut prédire comment une protéine se replie en trois dimensions, parfois avec une précision proche des expériences de laboratoire. Cette révolution ouvre la voie à la conception de nouveaux médicaments, enzymes et traitements sur-mesure.

Proteins, molecules ubiquitous in living organisms, are composed of a long sequence of small units called amino acids. These sequences follow hidden rules, like a biological grammar linking amino acids to the final shape of the molecule. Language models, complex Artificial Intelligence algorithms initially created to understand human text, have been trained on millions of protein sequences to decipher their secrets. Thanks to these models, AI can predict how a protein folds in three dimensions, sometimes with precision close to that of laboratory experiments. This revolution paves the way for the design of new drugs, enzymes, and tailored treatments.
4:00 pm to 5:00 pm

Digital twin and augmented surgery for bones and joints.
4:00 pm to 6:00 pm
  • Demo: "Human emotion understanding with AI" by Seongro Yoon (3IA Ph.D student, Chair of François Bremond)

This presentation marks the beginning of Seongro's Ph.D. journey under the 3IA program and introduces the research project I will pursue over the next three years: Comprehensive Emotion Understanding and Applications. The goal of this project is to explore and develop computational approaches for analyzing human emotional and psychological states using artificial intelligence. In particular, the research will investigate how multi-modal signals—such as facial video (computer vision) and physiological responses (e.g., EEG)—can be integrated to enable robust and context-aware emotion recognition. He will outline the motivations behind this work, its potential applications in AI-driven healthcare and human-computer interaction, and the technical roadmap for building models that go beyond surface-level cues to capture the deeper dynamics of human affect.