|
Education
Web
Viewing 1-3 of 3 total results
Science Podcast: Accurate Automatic Face Recognition; Complete Synthesis of a Genome; Unde...
Interviewee - Rob Jenkins This is very surprising to us. Once we fed the computer the average images, its performance was 100%, so it got all of the faces right. On standard photographs, when we were just showing it new photos of individuals who we...
1
0
Interviewee - Rob Jenkins This is very surprising to us. Once we fed the computer the average images, its performance was 100%, so it got all of the faces right. On standard photographs, when we were just showing it new photos of individuals who we knew were in the database, they got 54% of those right. We used an online implementation of an industry-standard face recognition system. This is at a fantastic website called myheritage.com. And it started out as a genealogy research website, but they kept
1
0
http://www.sciencemag.org/cgi/data/319/5862/499b/DC1/1#page=7
www.sciencemag.org/cgi/data/319/5862/499b/DC1/1#page=7
facial <span class="highlight">recognition</span> <span class="highlight">to</span> 100% by using what they call an "averaged" <span class="highlight">face</span>. In <span class="highlight">the</span> future, it could mean <span class="highlight">a</span> <span class="highlight">new</span> kind of photo for your passport. I spoke <span class="highlight">to</span> Jenkins from his office at <span class="highlight">the</span> University of Glasgow. Interviewee - Rob Jenkins As you know, there’s increasing interest in security applications and security infrastructure and biometrics in general. <span class="highlight">The</span> problem is, <span class="highlight">a</span> lot of <span class="highlight">the</span> <span class="highlight">face</span> <span class="highlight">recognition</span> systems that are out there simply don’t work at anything like <span class="highlight">the</span> level that people think they do. So for some
2
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
Interviewee - Rob Jenkins This is very surprising <span class="highlight">to</span> us. Once we fed <span class="highlight">the</span> computer <span class="highlight">the</span> average images, its <span class="highlight">performance</span> was 100%, so it got all of <span class="highlight">the</span> faces right. <span class="highlight">On</span> standard photographs, when we were just showing it <span class="highlight">new</span> photos of individuals who we knew were in <span class="highlight">the</span> database, they got 54% of those right. We used an online implementation of an industry-standard <span class="highlight">face</span> <span class="highlight">recognition</span> system. This is at <span class="highlight">a</span> fantastic website called myheritage.com. And it started out as <span class="highlight">a</span> genealogy research website, but they kept
Now Just a Blinkin' Picosecond!
provides a short-term solution to the problem. There are also physical problems accompanied by miniaturization that might affect the computer's reliability. " Drs. Frazier and Abdeldayem and their group in Huntsville, AL, have designed and built all-optical logic gat...
 Smithsonian: The Plant Press Newsletter Volume 4.2
difficulty with a system based on evolutionary lineages is its unfamiliarity. There is the sneaky suspi- cion that the names would vary freely with the current understandings of relation- ships, thereby losing the pretense of stability. Another possib...
1
0
difficulty with a system based on evolutionary lineages is its unfamiliarity. There is the sneaky suspi- cion that the names would vary freely with the current understandings of relation- ships, thereby losing the pretense of stability. Another possible difficulty is the uncertainty about how the new system would work for information recall, as is expected in the extensive files required for floristic studies. I could listen seriously to proposals for a more rational and objective taxonomic system, provided that I
11
0
http://botany.si.edu/pubs/plantpress/vol4no2.pdf#page=11
botany.si.edu/pubs/plantpress/vol4no2.pdf#page=11
difficulty with <span class="highlight">a</span> system <span class="highlight">based</span> <span class="highlight">on</span> evolutionary lineages is its unfamiliarity. There is <span class="highlight">the</span> sneaky suspi- cion that <span class="highlight">the</span> names would vary freely with <span class="highlight">the</span> current understandings of relation- ships, thereby losing <span class="highlight">the</span> pretense of stability. Another possible difficulty is <span class="highlight">the</span> uncertainty about how <span class="highlight">the</span> <span class="highlight">new</span> system would work for information recall, as is expected in <span class="highlight">the</span> extensive files required for floristic studies. I could listen seriously <span class="highlight">to</span> proposals for <span class="highlight">a</span> more rational and objective taxonomic system, provided that I
|
