Yes, there are many systems that do object classification from photos.
For example, this recent work of Poggio et al seems to me significantly more impressive than Hawkins work-so-far, cbcl.*mit*.edu/projects/cbcl/publications/ps/serre-wolf-*poggio*-PAMI-07.pdf both in terms of biological realism and actual object classification performance. However, it only tries to simulate the feedforward classification activity in the visual cortex, not the feedback connections. They plan to add biologically realistic feedback connections to their network in a later version. -- Ben On 5/14/07, Mike Tintner <[EMAIL PROTECTED]> wrote:
Thanks, Ben. If anyone knows more, I'd be grateful - that's such a huge, fundamental claim - basically "no computer can [up till now] recognize a cat, or any object" , that I think it's important to establish the historical truth of it. (He actually didn't distinguish between drawings/fotos. I did). Are you saying, for example, some current systems can recognize basic objects like cups/cats from photos? I guess that means: recognize "it's a cat" from either a) photos of several different cats and/ or b) photos of the same cat in several different positions. ----- Original Message ----- *From:* Benjamin Goertzel <[EMAIL PROTECTED]> *To:* [email protected] *Sent:* Monday, May 14, 2007 3:58 AM *Subject:* Re: [agi] Quick Hawkins Questions On 5/13/07, Mike Tintner <[EMAIL PROTECTED]> wrote: > > Here's a link to a lecture of his that's clearer than anything I've > read (incl. the book): > > http://mitworld.mit.edu/video/316/ > > Bottom line: his system can recognize simple objects from outline > drawings - like "dog", "cup." That seems to be the only concrete claim he's > making right now. There's no indication it can yet do anything more - like, > say. recognize simple movements ("dog bite", "cup fall") or recognize > objects from more complex photos Am I correct here? > > Secondly: he seems to be claiming that NO visual object recognition > system has achieved this before. If so, is he correct? > > > I'm not sure how many people have **tried** to make software systems that recognize outline drawings... Most vision researchers have focused more on recognition of images in photos... Probably if other researchers tried to tune their systems to recognize outline drawings, they could... -- Ben G ------------------------------ This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?&; ------------------------------ No virus found in this incoming message. Checked by AVG Free Edition. Version: 7.5.467 / Virus Database: 269.7.0/803 - Release Date: 13/05/2007 12:17 ------------------------------ This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?&;
----- This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?member_id=231415&user_secret=fabd7936
