Human Action Recognition and Monitoring in Ambient Assisted Living Environments

Magnani, Antonio (2020) Human Action Recognition and Monitoring in Ambient Assisted Living Environments, [Dissertation thesis], Alma Mater Studiorum Università di Bologna. Dottorato di ricerca in Computer science and engineering, 32 Ciclo. DOI 10.6092/unibo/amsdottorato/9373.
Documenti full-text disponibili:
[img] Documento PDF (English) - Richiede un lettore di PDF come Xpdf o Adobe Acrobat Reader
Disponibile con Licenza: Salvo eventuali più ampie autorizzazioni dell'autore, la tesi può essere liberamente consultata e può essere effettuato il salvataggio e la stampa di una copia per fini strettamente personali di studio, di ricerca e di insegnamento, con espresso divieto di qualunque utilizzo direttamente o indirettamente commerciale. Ogni altro diritto sul materiale è riservato.
Download (11MB)


Population ageing is set to become one of the most significant challenges of the 21st century, with implications for almost all sectors of society. Especially in developed countries, governments should immediately implement policies and solutions to facilitate the needs of an increasingly older population. Ambient Intelligence (AmI) and in particular the area of Ambient Assisted Living (AAL) offer a feasible response, allowing the creation of human-centric smart environments that are sensitive and responsive to the needs and behaviours of the user. In such a scenario, understand what a human being is doing, if and how he/she is interacting with specific objects, or whether abnormal situations are occurring is critical. This thesis is focused on two related research areas of AAL: the development of innovative vision-based techniques for human action recognition and the remote monitoring of users behaviour in smart environments. The former topic is addressed through different approaches based on data extracted from RGB-D sensors. A first algorithm exploiting skeleton joints orientations is proposed. This approach is extended through a multi-modal strategy that includes the RGB channel to define a number of temporal images, capable of describing the time evolution of actions. Finally, the concept of template co-updating concerning action recognition is introduced. Indeed, exploiting different data categories (e.g., skeleton and RGB information) improve the effectiveness of template updating through co-updating techniques. The action recognition algorithms have been evaluated on CAD-60 and CAD-120, achieving results comparable with the state-of-the-art. Moreover, due to the lack of datasets including skeleton joints orientations, a new benchmark named Office Activity Dataset has been internally acquired and released. Regarding the second topic addressed, the goal is to provide a detailed implementation strategy concerning a generic Internet of Things monitoring platform that could be used for checking users' behaviour in AmI/AAL contexts.

Tipologia del documento
Tesi di dottorato
Magnani, Antonio
Dottorato di ricerca
Settore disciplinare
Settore concorsuale
Parole chiave
Ambient Intelligence, Ambient Assisted Living, Human Activity Recognition, Computer Vision, Internet of Things
Data di discussione
3 Aprile 2020

Altri metadati

Statistica sui download

Gestione del documento: Visualizza la tesi