BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Modal Identification via Computer Vision - Mario Visconte\, PhD st
 udent\, University of Calabria\, UNICAL
DTSTART:20260220T160000Z
DTEND:20260220T170000Z
UID:TALK243166@talks.cam.ac.uk
CONTACT:46601
DESCRIPTION:Abstract:\nIn the last few years\, thanks to progress in compu
 tational power\, optical methods have found extensive application in the f
 ield of structural dynamics\; the main advantages reside in high-resolutio
 n or even full-field measurements of displacement or velocity\, enabling t
 he capture of modal vibration patterns of the structure\, and\, since they
  entail non-contact data acquisition\, in the ability to test under operat
 ional conditions where conventional sensors are difficult to deploy. Howev
 er\, current implementations still face limitations due to measurement noi
 se\, illumination variability\, and camera-motion artifacts\, as well as t
 he computational cost of high-resolution processing\, leaving room for sig
 nificant improvements in robustness and efficiency.\n\nThis talk will pres
 ent state-of-the-art computer vision approaches and our recent work to est
 imate resonant frequencies and extract operational deflection shapes via V
 ideo Motion Magnification (VMM). The proposed methods\, developed in part 
 during my research visit to the University of Cambridge\, are based on Dyn
 amic Mode Decomposition and a deep learning VMM model combined with signal
  processing\; they represent a step forward in Modal Identification via Co
 mputer Vision\, although some limitations and open questions concerning re
 liability in challenging conditions remain to be addressed.
LOCATION:JDB Seminar Room\, CUED
END:VEVENT
END:VCALENDAR