BEGIN:VCALENDAR
VERSION:2.0
PRODID:IEEE vTools.Events//EN
CALSCALE:GREGORIAN
BEGIN:VTIMEZONE
TZID:Europe/Rome
BEGIN:DAYLIGHT
DTSTART:20180325T030000
TZOFFSETFROM:+0100
TZOFFSETTO:+0200
RRULE:FREQ=YEARLY;BYDAY=-1SU;BYMONTH=3
TZNAME:CEST
END:DAYLIGHT
BEGIN:STANDARD
DTSTART:20181028T020000
TZOFFSETFROM:+0200
TZOFFSETTO:+0100
RRULE:FREQ=YEARLY;BYDAY=-1SU;BYMONTH=10
TZNAME:CET
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
DTSTAMP:20180928T163800Z
UID:484DE2FE-5C4F-4FFA-A871-F539A92FA3C1
DTSTART;TZID=Europe/Rome:20181019T100000
DTEND;TZID=Europe/Rome:20181019T110000
DESCRIPTION:In this presentation we will focus on the design of biometric c
 lassifier for forensic applications. Two examples will be addressed: foren
 sic face recognition based on facial features\, and face recognition at a 
 distance.\n\nExample 1: How Pattern Recognition Can Support the Forensic E
 xaminer\n\nIn this presentation we will show how properly designed classif
 iers can support the task of a automatic method to compare facial marks an
 d on how the result can be expressed in a likelihood ratio in order to qua
 ntify the evidential value. In addition\, we will illustrate to what exten
 d forensic facial comparison is different from biometric facial comparison
  as used in for instance access control.\n\nExample 2: Advances in Face-Re
 cognition at a Distance\n\nWe address the challenge of facial recognition 
 for surveillance applications. The typical problem here is the comparison 
 of a high-resolution reference image\, for example a mugshot\, with a low-
 resolution trace image taken at some distance\, for example found on a sur
 veillance video. I will demonstrate that realistic low-resolution images t
 hat are taken at a distance\, are not equivalent to low-resolution images 
 obtained by down-sampling higher-resolution images. This implies that in o
 rder to improve the recognition performance specific training of classifie
 rs is required\, but also that a proper evaluation on realistic low-resolu
 tion images is crucial. In the presentation\, I will discuss the implicati
 ons the on design\, training and testing of face recognition systems for s
 urveillance applications and propose a mixed-resolution classifier for thi
 s purpose. Attention will be paid to the deployment of convolutional neura
 l net based facial recognition systems for mixed-resolutions.\n\nSpeaker(s
 ): R.N.J. Veldhuis (Raymond)\, \n\nRoom: Multimedia room\, Bldg: B\, Roma 
 Tre University\, Dept. of Engineering\, Section of Applied Electronics\, v
 ia Vito Volterra 62\, Roma\, Lazio\, Italy\, 00146
LOCATION:Room: Multimedia room\, Bldg: B\, Roma Tre University\, Dept. of E
 ngineering\, Section of Applied Electronics\, via Vito Volterra 62\, Roma\
 , Lazio\, Italy\, 00146
ORGANIZER:patrizio.campisi@uniroma3.it
SEQUENCE:7
SUMMARY:Examples of the application of biometrics in forensics
URL;VALUE=URI:https://events.vtools.ieee.org/m/178152
X-ALT-DESC:Description: <br /><p>In this presentation we will focus on the 
 design of biometric classifier for forensic applications. Two examples wil
 l be addressed: forensic face recognition based on facial features\, and f
 ace recognition at a distance.</p>\n<p>Example 1: How Pattern Recognition 
 Can Support the Forensic Examiner</p>\n<p>In this presentation we will sho
 w how properly designed classifiers can support the task of a automatic me
 thod to compare facial marks and on how the result can be expressed in a l
 ikelihood ratio in order to quantify the evidential value. In addition\, w
 e will illustrate to what extend forensic facial comparison is different f
 rom biometric facial comparison as used in for instance access control.</p
 >\n<p>Example 2: Advances in Face-Recognition at a Distance</p>\n<p>We add
 ress the challenge of facial recognition for surveillance applications. Th
 e typical problem here is the comparison of a high-resolution reference im
 age\, for example a mugshot\, with a low-resolution trace image taken at s
 ome distance\, for example found on a surveillance video. I will demonstra
 te that realistic low-resolution images that are taken at a distance\, are
  not equivalent to low-resolution images obtained by down-sampling higher-
 resolution images. This implies that in order to improve the recognition p
 erformance specific training of classifiers is required\, but also that a 
 proper evaluation on realistic low-resolution images is crucial. In the pr
 esentation\, I will discuss the implications the on design\, training and 
 testing of face recognition systems for surveillance applications and prop
 ose a mixed-resolution classifier for this purpose. Attention will be paid
  to the deployment of convolutional neural net based facial recognition sy
 stems for mixed-resolutions.</p>
END:VEVENT
END:VCALENDAR