VOXReality

VOXReality is an ambitious project whose goal will be to facilitate and exploit the convergence of two important technologies, natural language processing (NLP) and computer vision (CV). Both technologies are experiencing a huge performance increase due to the emergence of data-driven methods, specifically machine learning (ML) and artificial intelligence (AI). On the one hand, CV/ML are driving the extended reality (XR) revolution beyond what was possible up to now, and, on the other, speech-based interfaces and text-based content understanding are revolutionizing human-machine and human-human interaction. VOXReality will employ an economical approach to combine these two. VOXReality will pursue the integration of language- and vision-based AI models with either unidirectional or bidirectional exchanges between the two modalities. Vision systems drive both AR and VR, while language understanding adds a natural way for humans to interact with the back-ends of XR systems or create multimodal XR experiences combining vision and sound. The results of the project will be twofold: 1) a set of pretrained next-generation XR models combining, in various levels and ways, language and vision AI and enabling richer, more natural immersive experiences that are expected to boost XR adoption, and 2) a set of applications using these models to demonstrate innovations in various sectors. The above technologies will be validated through three use cases: 1) Personal Assistants that are an emerging type of digital technology that seeks to support humans in their daily tasks, with their core functionalities related to human-to-machine interaction; 2) Virtual Conferences that are completely hosted and run online, typically using a virtual conferencing platform that sets up a shared virtual environment, allowing their attendees to view or participate from anywhere in the world; 3) Theaters where VOXReality will combine language translation, audiovisual user associations and AR VFX triggered by predetermined speech.

Our team was tasked with: 1) The development of Vision-Language models for spatially-aware image captioning and question-answering. 2) Transformers-based AI models optimization.

Consortium

MAGGIOLI SPA
ETHNIKO KENTRO EREVNAS KAI TECHNOLOGIKIS ANAPTYXIS
UNIVERSITEIT MAASTRICHT
STICHTING NEDERLANDSE WETENSCHAPPELIJK ONDERZOEK INSTITUTEN
SYNELIXIS LYSEIS PLIROFORIKIS AUTOMATISMOU & TILEPIKOINONION ANONIMI ETAIRIA
STICHTING DUTCH VIRTUAL REALITY DAYS
ADAPT IT AE
F6S NETWORK IRELAND LIMITED
HOLO-INDUSTRIE 4.0 SOFTWARE GMBH
ELLINIKO FESTIVAL ANONYMOS ETAIREIA