Multimodal Elder Care - MEC

The project intends to study computer vision and machine learning methodologies regarding the analysis of the pose, movements and behavior of people starting from multi-modal data. The application context regards the monitoring of bedridden patients with limited motor and cognitive abilities. In particular, the activities concern the study and development of vision algorithms for 3D modeling of the environment and for estimating the respiratory rate, and more in general the study and development of algorithms for the detection of anomalies in patient behavior. In addition, we intend to explore the ability to interpret situations with a model of Artificial Intelligence based on language (LLM) and language+images. This study also involves the development of tools for creating a dataset with related annotations.