Seminario “Computer Vision in Google Goggles”
Copiamos de la web del Coordinador del Grado la siguiente información:
Los estudiantes de Ingeniería Informática (de los estudios de cinco años y del grado) están especialmente invitados a asistir a este seminario.
Fecha: Lunes 14 de Noviembre de 2011
Hora:12:00
Duración: 1 hora
Lugar: Salón de Actos, Edificio Ada ByronTítulo: Computer Vision in Google Goggles
Ponente: Mark Cummins, Engineer, Google Inc.Abstract: Google Goggles is Google’s search-by-image system that aims to provide relevant search results given only a photograph of an object as input. Our long term aim is to build a universal visual search engine, capable of providing relevant information for any type of user photograph. While some categories of image are currently out of reach (e.g. plants, animals, weakly textured objects), the Goggles system is very successful in many categories such as artworks, tourist landmarks, manufactured products, books, wine labels, printed documents, etc. The primary technical challenge involved is to scale image retrieval methods to handle collections of tens to hundreds of millions of images. The talk will cover our findings in this area, as well as our ongoing efforts to expand the set of categories our system can recognize. In particular I will cover some of our recent results on reading text in uncontrolled user photographs, a neglected computer vision problem in which we have recently made progress.
Reseña biográfica del ponente: Mark Cummins obtained his PhD in Engineering working at the Mobile Robotics Group, University of Oxford, under the supervision of Paul Newman. He co-founded PlinkArt, which developed a mobile phone application designed to identify works of art in photographs. PlinkArt was acquired by Google in April 2010, the first UK acquisition of Google. Since then, Mark is part of the Google Goggles team (http://www.google.com/mobile/goggles/#text) and works in the improvement of visual search.