Top Message
Top Message
Back to Home Page  |  Recommend a Site  |  Settings   |  Sign In
Education Web
Target Your Results By: Clear All Selected iViews Customize Your Search
Viewing 1-1 of 1 total results
Science Podcast: Accurate Automatic Face Recognition; Complete Synthesis of a Genome; Unde...
we realized that we could exploit this setup to try to get at the performance level of that face recognition system on images of real world variability. So these are images that contain all the variability in lighting and so on that we are used to dealing with in the rea...
1 0
we realized that we could exploit this setup to try to get at the performance level of that face recognition system on images of real world variability. So these are images that contain all the variability in lighting and so on that we are used to dealing with in the real world. But it’s exactly that information that usually trips up the automatic systems. And so the logic behind this was that if the system is performing well, and we feed it a photograph of Bill Clinton, it should tell us that the closest
1 0 http://www.sciencemag.org/cgi/data/319/5862/499b/DC1/1#page=8 www.sciencemag.org/cgi/data/319/5862/499b/DC1/1#page=8
we realized that we could exploit this setup to try to get at the performance level <span class="highlight">of</span> that <span class="highlight">face</span> <span class="highlight">recognition</span> system on images <span class="highlight">of</span> real world variability. So these are images that contain all the variability in lighting and so on that we are used to dealing with in the real world. But it&rsquo;s exactly that <span class="highlight">information</span> that usually trips up the automatic systems. And so the logic behind this was that if the system is performing well, and we feed it a photograph <span class="highlight">of</span> Bill Clinton, it should tell us that the closest