BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Towards a User-Centric In-Vehicle Navigational System - Marwa Mahm
 oud (University of Cambridge)
DTSTART:20161020T131500Z
DTEND:20161020T141500Z
UID:TALK68523@talks.cam.ac.uk
CONTACT:Gyuri Denes
DESCRIPTION:Current navigational systems rarely consider generic road\nlan
 dmarks in their navigation instructions\, which can lead to\nmistakes\, fr
 ustration\, and distraction. However\, automatic detection of road landmar
 ks is difficult\, as current approaches to object detection focus either o
 n out-of-context objects which have special characteristics or on very spe
 cific domains. This work presents a future direction for a user-friendly n
 avigational system based on state-of-the-art computer vision techniques th
 at use deep learning for object detection. We propose an automatic hierarc
 hical approach for detecting and classifying a set of static and dynamic r
 oad landmarks that would be useful in automatic navigational systems. We f
 urther demonstrate a set of optimisations that improve performance and acc
 uracy of the basic system. We evaluate our approach on a natural\, ‘in-t
 he-wild’ dataset to determine how well it handles natural automotive inp
 ut. Finally\, we demonstrate a use-case for our system that extracts infor
 mation about a vehicle’s location and intention.
LOCATION:SS03 Meeting Room\, Computer Laboratory
END:VEVENT
END:VCALENDAR