Description du projet
Recourir à ’IA pour surveiller le trafic en temps réel
L’urbanisation rapide et l’augmentation permanente du nombre de véhicules ont suscité des inquiétudes concernant la sécurité routière. En conséquence, des systèmes de surveillance du trafic en temps réel ont été installés pour aider les opérateurs à contrôler la circulation et à faire face aux situations d’urgence. Le projet VISIONS, financé par l’UE, développera un système de surveillance du trafic en temps réel avec une transmission de haute qualité reposant sur des méthodes d’IA pour le traitement et la diffusion en continu de vidéo à destination des villes intelligentes. Les caractéristiques des systèmes visuels humains seront indiquées sur l’allocation de la qualité vidéo afin de limiter la portée nécessaire à la transmission vidéo. Une méthode avancée basée sur des réseaux neuronaux profonds sera déployée pour permettre le rendu et la diffusion en continu de vidéos à des résolutions plus faibles, et un nouveau système d’adaptation du débit binaire basé sur l’apprentissage par renforcement sera prévu pour garantir la qualité de l’expérience.
Objectif
With the rapid development of urbanization and continuous increase of vehicles on roadways, Intelligent Transportation Systems (ITS) play a key role in revolutionizing the way people commute. To make our cities safer and smarter, real-time traffic monitoring systems are deployed to help operators with observing traffic flows and identifying emergency situations.
This project aims to achieve real-time traffic monitoring with high-quality video transmission in smart cities, leveraging the emerging Artificial Intelligence methods in video processing and video streaming. Firstly, the features of human visual systems will be referred on video quality allocation to reduce the required bandwidth for video transmission. Next, an innovative method for end-to-end video processing based on Deep Neural Networks will be developed to allow the video rendering and streaming at a lower resolution and also restore/improve the quality at the user ends. Finally, a new bitrate adaption scheme based on Reinforcement Learning will be designed to accommodate the unexpected network dynamics, guaranteeing the Quality-of-Experience to be perceived by users. The expected outcome can promote safer and more efficient travel for millions of users in Europe and billions of users all over the world. Moreover, the results of this project can be used in other multimedia applications, such as cloud virtual reality, distance education, smart transportation, and healthcare where video processing and video streaming are needed.
To broaden the fellow’s knowledge horizon, a series of research, training, and knowledge transfer activities are planned. The new knowledge and skills imparted in these activities will further promote his academic portfolio and significantly enhance his career prosperity. The project will also play a solid foundation for the long-term and wide-range collaborations and eventually lead to more extensive impact of project results, from which both EU and China will benefit.
Champ scientifique
- engineering and technologycivil engineeringurban engineeringsmart cities
- natural sciencescomputer and information sciencesartificial intelligencemachine learningreinforcement learning
- natural sciencescomputer and information sciencessoftwaresoftware applicationsvirtual reality
- natural sciencescomputer and information sciencesartificial intelligencecomputational intelligence
Programme(s)
Régime de financement
MSCA-IF - Marie Skłodowska-Curie Individual Fellowships (IF)Coordinateur
EX4 4QJ Exeter
Royaume-Uni