Share this page

Start-up projects funded by the "3IA Start-it-up program"

Projects in progress

DAD ECG by Inn’Pulse AI project for Digital Automatic Defibrillator Electrocardiogram Chair: Maxime Sermesant - Inria (Epione team)

In France, 50,000 cardiac arrests occur every year, of which only 5% are recovered. The main cause is a sudden acceleration in heart rate. After cardiac arrest, every minute without treatment reduces the chances of survival by 7 to 10%. It is therefore essential to quickly alert, massage and defibrillate.

Today's defibrillators are rarely accessible and difficult to use. 80% of cardiac arrests occur at home.
Based on these observations, Inn'Pulse aims to facilitate access to defibrillation and save lives, with a miniature defibrillator that is accessible (including at home and in any location), easy to use and maintenance-free.

The project led by Maxime Sermesant, holder of a 3IA Chair, involves the development of a new computer program using Artificial Intelligence to analyze cardiac signals. This program must be capable of accurately detecting heart rhythms, while effectively eliminating parasites and interference that can affect the quality of recorded signals. It is designed to be integrated into the Inn'Pulse miniature defibrillator controller, whose other modules will control data recording and electrical impulses, as required.

Target markets include medical and paramedical professionals working in the home, first aiders in the workplace, firefighters, internal security forces and volunteer first aiders.

About Maxime Sermesant:
He is a permanent researcher at Inria. He is Head of Computational Cardiology at Inria Epione and Head of Multimodal Data Science at IHU Liryc. He is also the co-founder and scientific advisor of inHEART and co-founder of Therapixel. His research interests include biomedical image processing, organ modelling and machine learning.

Harness Vision AI image processing for horse well-being Chair: François Bremond – Inria (STARS team)

The Harness Vision project is based on a prototype already developed at the Mougins equestrian center by Harness Vision. Industrialization began in March 2023, following incubation at the PCA incubator. The first version of the solution (Minimum Viable Product or MVP) is due to go on sale in April 2024. The second version, with all functions for professionals (veterinarians in particular), especially for disease diagnosis, is due for delivery in April 2025.

Led by François Bremond, holder of 3IA Chair, this scientific collaboration with the 3IA Côte d'Azur Institute has two main objectives:

Integrate the latest AI technologies into image processing;
Use the Institute's experience in other fields (image processing for human behavior analysis in particular) to apply it to horses.

The target market is "premium" riding schools and racehorse training stables, i.e. around 70,000 horses in France. The commercial strategy is to swiftly expand into other European markets (Germany, Benelux, UK), each offering comparable or larger market sizes. The horse racing and sales sector is already international and faces fewer regulatory hurdles than the human health market.

About François Bremond:
He is a Research Director DR1 at INRIA Sophia Antipolis. He created the STARS team in 2012. In 2007 he obtained his HDR degree (Habilitation à Diriger des Recherches) from Nice University on Scene Understanding: perception, multi-sensor fusion, spatio-temporal reasoning and activity recognition. He is a co-fonder of Keeneo, Ekinnox and Neosensys, three companies in intelligent video monitoring and business intelligence. He also co-founded the CoBTek team from Nice University on the 1st of January 2012 with P. Robert from Nice Hospital on the study of behavioral disorders for older adults suffering from dementia.

L.I.M.P.I.D. (Leveraging Images for Modelisation and rePresentation In 3D) by Videtics Use RGB images describing a scene to model and represent manually detected and added objects in 3D space Chair: Charles Bouveyron – Inria / CNRS / Université Côte d’Azur (MAASAI team)

The overall aim of the project led by Videtics, a company incubated by the PACA Est Incubator, is to implement representation methods that provide a better understanding of the space of the observed scene and the objects within it.
Generally speaking, the aim is to understand when a change in the observed scene takes place, and to propose a mapping between 3D representation and video scene.

Led by Frédéric Precioso (a lecturer at Université Côte d'Azur and Polytech Nice Sophia, and a researcher in the MAASAI team), the project aims to analyze video scenes. This objective is directly linked to the MAASAI team's current research projects on autonomous vehicles and biodiversity through the analysis of nature scenes.

Initially, Videtics designed software integrating its own detection solutions and object tracking algorithms to create a modular package. The company then developed a range of solutions specifically adapted to the target markets:

Intelligent urban video in cities, ports and transport;
Perimeter protection and logistics for private companies and sensitive sites;
Perimeter video protection for prestigious villas and yachts.

About Charles Bouveyron:
Charles Bouveyron is the Director of 3IA Côte d'Azur. He holds a Chair on Artificial Intelligence and he is full Professor of Statistics (Professeur des Universités) at Université Côte d'Azur. He is also the head of the research team MAASAI on Statistical Learning and Artificial Intelligence, which is a joint team of INRIA Sophia-Antipolis and Université Côte d'Azur. He serves as an associate editor for The Annals of Applied Statistics and he is the founding organizer of the series of Statlearn workshops.

NeuroDec Producer of AI-based algorithms for simulating and analyzing muscle signals Chair: Rachid Deriche – Inria (ATHENA team)

Neurodec develops artificial intelligence algorithms for advanced processing of non-invasively recorded muscle signals. These algorithms can be used in:

Robotic control (target markets: bionic prostheses/orthoses, exoskeletons, etc.);
Neurofeedback (target market: rehabilitation);
Diagnostics (target market: treatment of muscular diseases and disorders);
Interaction with virtual intoxication (target market: virtual reality).

Led by Samuel Deslauriers-Gauthier (Inria, ATHENA team), the project is developing a "myoelectric prosthesis" technology, which uses AI algorithms to translate muscle signals into prosthesis movements.

The start-up's "go-to-market" product is a realistic electromyogram (EMG) signal simulator that can be used to generate large, realistic supervised databases for training algorithms. This cutting-edge technology will enable prosthesis manufacturers to increase the punctuality of prostheses, reduce production costs and shorten time-to-market for new products. It will also make the use of prostheses more efficient, intuitive and easy, greatly improving the quality of life of amputees.

About Rachid Deriche:
Rachid Deriche is a research director at Inria Sophia Antipolis, where he leads the research project Athena aiming to explore the Central Nervous System using computational imaging. He has made major contributions to the scientific community, mainly in image processing, computer vision and neuro-imaging. In 2019, he received the title "EURASIP Fellow", the most prestigous honour from the European Association for Signal Processing (EURASIP). In 2016, Rachid Deriche received from the European Research Council (ERC) a grant to support his research project “Computational Brain Connectivity Mapping”.He also received the title of Honorary Doctor (honoris causa) from the University of Sherbrooke in 2014, to honor his scientific contributions and his approach in educating new generations of scientists. In 2013, he was awarded Grand Prize of the EADS Corporate Foundation in Computer Science by the French Academy of Sciences for all his outstanding scientific contributions in computer science.

NeuroPin AI-based software for augmented neuroradiology, which improves the quantification and tracking of neurological change over time from neuroimages Chair: Marco Lorenzi – Inria / Université Côte d’Azur

NeuroPin’s ultimate purpose and objectives are to improve lifelong radiology by improve the quantitative basis of radiological interpretation and diagnostics.

The start-up develops AI-powered software for the quantitative analysis of brain MRI scans, aimed at enhancing clinical diagnosis and treatment planning. This can be used for:

Brain segmentation + mapping
Target market: educational purposes, radiological collaboration, research

Volumetry of brain regions and anomalies
Target market: hospitals, radiological centres

Brain charts: tracking change over time which can lead to:
> Improved understanding of longitudinal disease change
> Individualised medicine for lifelong care
> Population-level interpretation of surgery and/or treatment response.
Target market: pharmaceuticals, neurosurgery

Collaboration between NeuroPin team and Marco Lorenzi, holder of 3IA Chair, allows the start-up to gain knowledge and experience in order to train and test its model, as well as the guidance, feedback and advice from Dr Lorenzi who already has experience in fostering collaborations for empowering clinical data sharing and interpretation.

About Marco Lorenzi:
He is a tenured research scientist (CR) at Université Côte d’Azur, Inria Sophia Antipolis. His research interest is in the the development and study of computational and statistical methods for the analysis of biomedical data and brain images. His current research topics include Bayesian modeling and uncertainty quantification, time-series analysis, latent variable models, and federated learning.

RO3SE (RObust Semi-Supervised Speech Enhancement) by Pulse Audition Improving hearing comfort for persons with hearing loss thanks to AI Chair: Pierre-Alexandre Mattei – Inria (MAASAI team)

Pulse Audition is a French startup whose mission is to improve the social integration of people suffering from hearing loss. The Pulse Frames are the first AI-powered intelligent hearing glasses that give crystal-clear speech perception in noisy environments, to improve social connections. These glasses, developed by the Pulse Audition, are designed to make hearing loss imperceptible and meet the needs of mild-to-moderate hearing-impaired people, enabling them to hear only what they want to hear in noise. In France, this target market represents 12 million people.

The partnership between Pulse Audition and the MAASAI team, and more specifically with Pierre-Alexandre Mattei, holder of 3IA Chair, aims at improving the robustness of the real-time speech enhancement algorithms integrated into the Pulse Frames, by leveraging unsupervised/unlabeled data that correspond to real-life audio scenarios.

About Pierre-Alexandre Mattei:
He is a Research Scientist at Inria. He is part of the Maasai (Models and Algorithms for Artificial Intelligence) team and is also affiliated with the J.A. Dieudonné lab and holds a 3IA Côte d'Azur chair.
His field of research is statistical machine learning, with a particular emphasis on hidden variables and model uncertainty. He is a co-organiser of several annual events related to machine learning: Statlearn, GeMSS, GenU, and the SophI.A Summit.

SECUMAD (Solution Multi-capteurs Sécurisante de Maintien à Domicile pour Personnes Agées Vivant Seules : Secure Multi-sensor Homecare Solution for Seniors Living Alone) Digital solution dedicated to safe home care for elderly people living alone Chair: François Bremond – Inria (STARS team)

The elderly population is set to grow considerably over the next 20 years. The number of people requiring care, particularly those suffering from Alzheimer's disease, will increase accordingly, while the number of people able to provide suitable home care tends to decrease. This will lead to a significant loss of autonomy, forcing patients to move into specialized medical establishments, creating psychological problems and heavy financial burdens for both the patient's family and the healthcare system.

Led by François Bremond, holder of 3IA Chair, the SECUMAD project is a digital solution based on a set of IoT and video sensors, enabling the activities of daily living (ADLs) of the elderly to be monitored. It is the sequel to SOLITARIA, a joint project between the STARS team, the CoBTeK laboratory (Université Côte d'Azur / Inria) and the FS company. This collaboration has resulted in the development and validation of a set of technological bricks that will form the basis of the SECUMAD solution.

SECUMAD, which will be marketed under the name KoKoon by start-up Nodeus Solutions, is aimed at the Silver Economy market. It will be directly accessible to the elderly, even those on modest incomes.

About François Bremond:
He is a Research Director DR1 at INRIA Sophia Antipolis. He created the STARS team in 2012. In 2007 he obtained his HDR degree (Habilitation à Diriger des Recherches) from Nice University on Scene Understanding: perception, multi-sensor fusion, spatio-temporal reasoning and activity recognition. He is a co-fonder of Keeneo, Ekinnox and Neosensys, three companies in intelligent video monitoring and business intelligence. He also co-founded the CoBTek team from Nice University on the 1st of January 2012 with P. Robert from Nice Hospital on the study of behavioral disorders for older adults suffering from dementia.

SequoIA analytics DAS traffic data analysis for the smart city Chair: Cédric Richard – Université Côte d’Azur (Lagrange Laboratory)

The aim of the project is to develop a suite of tools for monitoring and diagnosing roadways and traffic infrastructures in the service of intelligent cities and territories. The unique feature of these tools, based on artificial intelligence, is that they can simultaneously process roadway usage and road geotechnics in real time.

Led by Cédric Richard, holder of 3IA Chair, SequoIA analytics offers an innovative monitoring technology based on Distributed Acoustic Sensing (DAS). This solution uses existing telecom optical fibers to detect seismo-acoustic disturbances over a distance of more than 100 kilometers, without the need for additional equipment in roadways. The system can track road, rail and pedestrian traffic along the entire length of the fiber in real time. The project stands out for its ability to transform complex and voluminous DAS data into interpretable information, offering a non-invasive and versatile solution for infrastructure monitoring.

The target market is infrastructure operators (Vinci, Véolia, etc.) and local authorities, who are constantly on the lookout for solutions to monitor their infrastructures (to track pavement deterioration, for example) or traffic (road traffic).

About Cédric Richard:
Cédric Richard is Professor at the Laboratoire Lagrange. He has been a member of the Institut Universitaire de France (IUF) in 2010-2015. His research interests lie at the intersection of statistical signal processing and machine learning, and how they have synergistic effects on system monitoring. In particular, his recent research works mostly focus on online system identification in a variety of forms, including: online learning, signal processing and learning over graphs and networks, adaptive signal processing, nonlinear systems and signal processing.

SPLEAT Exploiting the SPLEAT low-power neural processor Chair: Benoît Miramond – Université Côte d’Azur (LEAT : Electronics Antennas and Telecommunications Laboratory)

LEAT and the 3IA Côte d'Azur Institute, through Benoît Miramond's chair, have distinguished themselves in research into spiking neural networks (SNN), considered to be the third generation of artificial neural networks. These SNNs, inspired by the brain workings, are being studied for their energy efficiency, while retaining the principles of Deep Learning.

LEAT has developed SPLEAT (SPiking Low-power Event-based ArchiTecture), a specialized computing architecture for SNNs. In 2020, this innovation led to a world premiere: the autonomous use of a neuromorphic computer in space, as part of IRT Saint Exupéry's CIAR project.

Led by Edgar Lemaire (Université Côte d'Azur / CNRS – LEAT), the project aims to develop software and hardware solutions for deploying very low energy Deep Learning models close to the sensors directly on neural processors. This approach is part of a frugal, decentralized and autonomous AI approach, responding to today's energy and environmental challenges.

SPLEAT primarily targets the market of integrated circuit designers and manufacturers of autonomous embedded systems, offering its solutions in the form of Intellectual Property (IP) or integrated circuits, adaptable to various players in the emerging Edge AI market.

About Benoît Miramond:
He is a full Professor at the laboratory LEAT of Université Côte d'Azur. He is currently the head of the Neuromorphic Engineering Group from LEAT. He is currently the head of the eBRAIN group from LEAT. He participates to the EMERGENCES project from the french PEPR AI programme from 2023 to 2028.
In this context, his research is following an interdisciplinary approach to explore novel adaptive neural architectures inspired from neurosciences and cognitive sciences for Edge AI applications.

Past projects

GraphKey: Network and text analysis using AI algorithms
Chair: Charles Bouveyron – Inria / CNRS / Université Côte d’Azur (MAASAI team)

GraphKey draws on its catalog of original AI algorithms, including a proprietary technology. The aim of the project, led by Charles Bouveyron, holder of 3IA Chair, is to create a Tech company capable of answering the most advanced business questions in marketing, healthcare and cybersecurity using network-type data.

The catalog will contain several algorithms for network clustering (SBM), text classification (LDA, sentiment analysis, etc.) and simultaneous network clustering with text edges (Linkage).

Linkage is an innovative and unique technology for the simultaneous clustering of networks and text. It identifies hidden groups (clusters) within a network, while uncovering the main (unknown) themes discussed between them. This type of data source could be, for example, text exchanges between individuals in a social network such as Facebook or Twitter, e-mail exchanges between employees in a company, or "co-authors" of patents or scientific publications.

About Charles Bouveyron:
Charles Bouveyron is the Director of 3IA Côte d'Azur. He holds a Chair on Artificial Intelligence and he is full Professor of Statistics (Professeur des Universités) at Université Côte d'Azur. He is also the head of the research team MAASAI on Statistical Learning and Artificial Intelligence, which is a joint team of INRIA Sophia-Antipolis and Université Côte d'Azur. He serves as an associate editor for The Annals of Applied Statistics and he is the founding organizer of the series of Statlearn workshops.
RLCRS by Qiti: Maturation of a French-language conversational recommendation system for insurance retailing
Chair: Giovanni Neglia – Inria (NÉO team)

Qiti is an InsurTech company developing a virtual cognitive assistant (VCA) for policyholders and insurers, of which the warranty recommendation engine (RecSys) is one of the key functionalities.

The ambition of the project, led by Konstantin Avrachenkov (Inria, NÉO team), is to reconcile policyholders and insurers by providing an innovative, digital solution for recommending cover and insurance, so that policyholders have suitable, scalable cover over time, based on their needs, their specific situation and their environment.

The start-up was able to determine the causes of the tensions that arise during the retail insurance purchasing process and the life cycle of a contract, and to identify conversational recommendation system (CRS) technology as the best candidate for resolving these issues. Indeed, it relies on the AI agent's real-time updating and automatic interaction capabilities with a dynamic environment, enabling it to extend the portion of the environment observable by the AI agent and to refresh previous observations of this environment in order to improve the relevance of recommendations.

The LCA is intended to be exploited commercially directly by Qiti in the digital nomad market segment, and to be integrated into the systems of companies wishing to offer insurance recommendations to their customers.

About Giovanni Neglia:
Giovanni Neglia is a Researcher at Inria. His research interests include performance evaluation of distributed systems, in particular cache networks and large-scale learning systems. His research is characterized by the application of different mathematical tools (Markov processes, control theory, continuous optimization, fluid models, game theory).

Learn more about the Start-it-up program