[Hmm. lne.com spam-blocked me on the first attempt. Given the identity of the research group, I forgot to add the obvious "The computer says he's a rambling wreck from Georgia Tech." ]
At 01:36 AM 10/27/2002 -0400, Declan McCullagh wrote: >See also: >http://www.wired.com/news/politics/0,1283,38775,00.html BTW, The Wired news article is from 2000, and mainly focuses on facial recognition, but it's interesting that this stuff has been in the press for a while. It's wrong in so many different ways... "Why'd you administratively detain him in the airport?" "He walked like an Egyptian" "The Ministry of Silly Walks put out the following press release..." "Why'd you bust him?" "He was loitering suspiciously" "He was shuffling like known suspect Uncle Tom" "She had high-heeled shoes in her suitcase, obviously intending to disguise her gait for the cameras" "He had high-heeled shoes in _his_ suitcase, ..." There are potential medical uses of this sort of technology - enough computer abusers and other desk-job workers with bad backs or similar health problems that could benefit from analyzing how they walk, but obviously Darpa's not going to find that. Perhaps we can get some chiropractic foundation to redirect the study personnel, or get Darpa to use it to help design better backpacks for soldiers instead of using it is spookware. Or perhaps we're better off having them focus their studies on something highly unreliable like this instead of spookware that might be more usable. >Subject: NCP: Privacy Villain of the Week: DARPA's HumanID at a Distance > >The Defense Advance Research Projects Agency <http://www.darpa.mil/> has >been one of the more fruitful government agenies in the past, its DARPAnet >computer network being the foundation for what would become the Internet >some years later. That is why reading about what this outfit is up to now >can at times be disheartening. One such project is the HumanID at a >Distance program, which aims to move beyond face-recognition technology >and purportedly identify people by the way the walk. > >The idea here is that by measuring with video or (clothes-penetrating) >radar the distance between, say, 17 different points on the body ><http://gtresearchnews.gatech.edu/images/gait1.jpg> and measuring how >these points move in relation to one another, a person can be positively >and uniquely identified. ....
