Towards storytelling from visual lifelogging

22 Aprile, 2016
Versione stampabile

Time: 10:00am
Location: Meeting Room Garda - Polo scientifico e Tecnologico "Fabio Ferrari" (Edificio Povo 1, via Sommarive 5 – Povo, Trento)

Speaker:

  • Mariella Dimiccoli-Computer Vision Center, Universitat Autonoma de Barcelona 

Abstract:  
Visual lifelogging consists of acquiring images that capture the daily experiences of the user by wearing a camera over a long period of time. The collected data, commonly called visual lifelogs, offer considerable potential for inferring knowledge about a person’s life and hence enable many applications such as contrasting dementia by cognitive training based on digital memories and monitoring lifestyle for the prevention of noncommunicable diseases. However, visual lifelogs present a significant challenge for automatic visual analysis, mainly due to the free motion and to the low frame rate of the camera (up to 2 pictures per minute), and to the nonintentional nature of the images. Furthermore, the sheer number of data that a visual lifelog consists of and the rate at which they increase (up to 2,000 images per day) imposes a need for efficient methods to extract and locate relevant content. In this talk, I will discuss recent results (and some ongoing work) on the automatic analysis of visual lifelogging. In particular, I will focus on the problems of 1) event representation, 2) activity recognition and 3) visual question answering, for which the use of language processing would be highly beneficial, if not necessary. 

About the Speaker: 
Mariella Dimiccoli (http://www.ub.edu/cvub/marielladimiccoli/) received the MsC degree from the Polytechnic University of Bari in 2004 and a Master of Advanced Studies (MAS) degree in 2006 from the Technical University of Catalonia, where shecompleted a PhD in Image Processing in 2009. Afterwards, sheI spent one year as postdoctoral researcher in the Laboratory of Physiology of Perception and Action at Collège de France-CNRS (France) and more than two years in the Laboratory of Applied Mathematics (MAP5-UMR68145-CNRS) at Paris Descartes University. From 2013 to 2014 she was a Visiting Professor in the Image Processing Group (GPI) at University Pompeu Fabra (UPF). She joined the Barcelona Perceptual Computing Lab (BCNPCL) with a Beatriu de Pinós grant in October 2014. 

Contact persons regarding this talk: Nicu Sebe, niculae.sebe [at] unitn.it & Raffaella Bernardi, raffaella.bernardi [at] unitn.it