Slashdot Mirror
Analyzing (All of) Star Trek With Face Recognition
An anonymous reader writes "Accurate face recognition is coming. Pittsburgh Pattern Recognition, a face recognition start-up spun out from Carnegie Mellon University, has posted a tech demo showing an analysis of the entire original Star Trek series using face recognition. The online visualization includes various annotated clips of the series with clickable thumbnails of each character's appearance. They also have a separate page showing the full data of all the prominent characters in every episode including extracting thumbnails of each appearance." Their software can recognize frontal or near-frontal face instances.
know the name of that red shirted guy?
Unless you're looking pretty much straight-on towards the camera, this software doesn't appear to work. It does appear to be able to track a face over multiple frames if it can recognize it in one frame, but if you have 30 seconds where no suitable frame occurs, the software doesn't know who it is, even if it's pretty blatantly obvious to a human who it is.
paintball
What a wellspring of Trek trivia! For instance, in Spock's Brain, Spock gets less than 4 and a half minutes of screen time. Fascinating...
Shatner's a camera hog...
Todos mis movimientos están friamente calculados
Why is this in YRO? It's just plan cool.
A Magic the Gathering Article and Forum Aggregator
We forehead-challenged beings demand you stop your software discrimination!
About a year or two ago
What I wanted was for face recognition software to become more general so you could search for movies using vague memories from your childhood:
"Girl on boat", "Wheat field", "Yellow flag"
With an advanced enough search engine, you could tag everything automatically.
I didn't think of privacy concerns though, I guess thats a good point.
Are you serious? Not only is it a fantastic demonstration of face recognition technology under real world conditions, but it's also incredibly useful.. if you have a little vision. How many times have you been watching a tv show and said "wow, where have I seen that guy before?" To find out these days I typically do:
1. Take note of the show I'm watching and the episode name (if given).
2. Go to imdb and hope they have specific info on that specific episode.
3. Try to guess what the character's name was, and take note of the actor's name.
4. Click through to the actor's filmography.
And, most typically, one of those steps fails. Now imagine if your tivo or other media playing device had face recognition technology like this. You'd just press one button and it would put boxes around all the faces on-screen, you'd select the face you're interested in and it would immediately tell you the name of the actor, the name of the character that actor is playing in this episode, how many other episodes of this series that he's in, and the actor's entire filmography. That's a real product that I'd actually buy!
How we know is more important than what we know.
Jesus, that's a lot of work to go through to figure out that Bruce Campbell has been in a shitload of B-movies.
You wouldn't believe how many of the actors from Battlestar Galactica were in The Dead Zone.
How we know is more important than what we know.
Machine recognition (facial or otherwise) is a lot easier in the lab setting, where you have a small pool of objects to recognize and a lot of data on each object. In typical non-lab settings, you have a large pool of objects to recognize and a small amount of data on each object, and your recognition rates (fp and fn) go south pretty quickly.
Or popping up in Dollhouse now.
> UnbalancedSimpletons.com [com.com]
You were trying to reach www.unbalancedsimpletons.com.com, which doesn't exist.
Yeah I know, it's a bit of an inside joke.
I would hope so, but how is this not using someone else's copyrighted crap for commercial gain?
This is a fun demo of their product, and realistically can't affect Star Trek revenues in a negative way. However, I suspect that some Paramount copyright lawyer might be getting wood about now.
__ Someday, but not this morning, I'll finally learn to use the preview button.
If you let it play a few minutes you'll see it indenify Spock and then in the next scene he comes up as unknown even though he's facing the camera. The system seems to fail when he arches his eyebrows.
and nothing of value was ... wait, they did what?...
not only is time travel possible, it's irrelevant.
I am not sure how much I really trust that per-episode chart once I started looking at one of the sample episodes. At some points the face recognition will not pick up on main characters for apparently no reason. For example, in episode 53 at about 1:30 in the sample there is dialog between Spock, Kirk, and someone else. The camera angles are steady and consistent (other than people turning their head while talking) and sometimes the system does not recognize one of the characters after it did just a few seconds earlier. On the right side it shows the name of the character or "Other" if it recognizes a face, but not compared to something it knows about.
Overall, this is simply amazing!
"Mostly harmless."
Accurate face recogition is coming!!! Really? Imagine if a company that made photo software made it so that the software would recognize the faces in your photos and then would automatically find the same faces and let you assign a name to them. Then you could look at all the photos of that person without having to manually tag the photos! Just imagine if someone did that. I guess we'll have to wait for accurate face recognition to actually come. [/sarcasm]
Checkout http://facemining.pittpatt.com/S3E75/, Scotty shows up under Kirk twice, and thats with just one try.
Or, http://facemining.pittpatt.com/S1E12/ actor 0117 has an odd match on my second peek.
They might want to try shirt matching.
(3) probably most important: out of copyright.
Really?
Is recognize the faces of actors in makeup that may change their facial features. It doesn't recognize Vina when she appears in the 'non-illusion' state, although to real people she is still easily recognizable as Vina. So they have a 'ways to go' with their capabilities.
Sig this!
There is nothing in principle bad about what is being demonstrated here. But let's be clear, this isn't a new step forward, this quality of facial recognition has been around for years (just ask Vegas).
The biggest limitation on facial recognition is and has always been the data processing cost. In terms of that the technology is obviously getting more and more viable as hardware progresses.
Might we see this in a TV in the future? Maybe. But only when the cost of the hardware gets to a certain point and then someone successfully markets that to the general populace.
Which is scarier?
Approaching Cylons or your consciousness on a WD Green drive?
Help! Help! I'm being repressed!
Identifying poster IP...
Tracking User with GUI interface written in Visual Basic to track the IP address...
Sending kill pulse over Internet to disable user communications...
Isolating coordinates on power grid, and disabling power service...
Dispatching unmarked black vans and helicopters for potential terrorist pickup... Use of deadly force authorized...
When our agents reach your house, don't resist. It'll just make your death more painful, and make our agents work harder. The outcome is inevitable. On never mind, you can't see this, now can you. But maybe others will learn from your mistakes.
Have a nice day.
Serious? Seriousness is well above my pay grade.
If it were me, I'd use "ZERO ZERO ONE ZERO ONE ZERO ONE ZERO" in the ultimate sig!
This issue is a bit more complicated than you think.
Sure, do that for all the people entering a stadium ? Now spot the guy with a criminal record.
...
meanwhile, two years ago
http://www.hostcity.co.uk/features/stadium_tech/face.html
This is cool technology for my Tivo, but it also sounds like this technology could scan through the thousands of video feeds collected by municipal surveillance cameras and track my every movement. Not that my life is so interesting to anyone else, but what if it were?
To be fair though, facial recognition software doesn't work on Bruce Campbell. It recognizes most of his face correctly, but locks up every time it gets to his chin.
So this software can recognise a face shown in a screenshot or video clip. So what?
You could simply hold up a picture to the thing to fool it, or an iPhone with the first season of Star Trek playing it would seem.
What a waste of time. The only facial recognition worth mentioning is the pattern projection method (gah, can't find link), which actually requires you have a 3D face for it work, but even then you can always trick it
I'd heard it said recently that things that computers can do in the realm of facial recognition, speech-recognition, and (a little more obviously) optical character recognition have come a damned long way, far further than most realize. That most people's experience with, say, speech-recognition is through some free-ware (crapware actually) and they don't know just how good the state of the art is. Which is: damned good.
Speech-recognition is essentially a solved problem. OCR is easy, obviously, with all the CAPTCHA news going around. Face recognition, sure, sure, this article proves it. So, what's next?
Areas that computers do not excel yet: writing a review of a book or movie, inferred of information, etc. But, it's though that by about 2025 the number of transistors and speed of processors will be such as to rival the brain and after that point all bets are off. It will be an exciting 15 years in AI research.
You might think of a human as a program running in physical memory space. When the power shuts off, we die. But, if we imagine a conscious program we can imagine a being who can 'image' every moment of life (or of their brain), save it, and even rewind backwards, or stop and start states, easily. If you're an AI and you see something you don't want to remember, just rewind a bit and it's gone forever :P If we become: Techumans, that is, human intelligences uploaded into the machine, then life becomes far more interesting going forward on so many fronts...
"I Don't Have Enough Faith to be an Atheist"
Most of us have seen the pictures of children who have gone missing. These pictures appear at Walmart, in the 1040 publication from the IRS, etc. Many such pictures contain the faces that have been advanced in age by computer software.
However, the reach of such pictures is limited. Few people pay attention to such pictures. Of the people who care enough to notice, they are not constantly looking for the missing children in order to report them to the police.
How can this new face-recognition software help? Most restaurants (like McDonalds), most movie theatres, and the like already have cameras that film everyone entering and leaving the premises. The government should feed these image streams into a cluster of supercomputers owned by the FBI and running this new face-recognition software. It will then match faces (in the crowds) against the age-advanced images of the missing children. Such a supercomputers could run 24 hours for 7 days per week and scan images that are fed from millions of locations across the USA.
In such a scenario, the chances of finding the missing children would be greatly improved.
Given the pittpatt founders come from CMU, I'm sure they are brilliant and will find more creative and interesting ways to turn pattern recog into $$. However, is this tech really new/cutting edge? Facial recog has been around for a while. Heck, I had to write similar software as projects in grad school. Sounds like pittpatt founders might have developed faster, more accurate/reliable algorithms (far better than I could do). However, is this really enough to support a new company? Maybe the interesting details just haven't been made available. If not, then this may be a premature slashdotting. Comp vision, pattern recognition, and machine learning are all popular courses at top comp sci grad schools. Hence, I expect to see hordes of new companies and lots of competition for these folks in the not too distant future. Having a better algorithm is not nearly as good as having a great, new, innovative product.
If you don't want to be recognized by facial recognition software, wear black and white greasepaint. In Episode 70 the actors playing Lokai and Bele are misidentified in a few scenes. The images are categorized by the black and white makeup rather than the actor. I'm not sure why Bele's face paint is reversed in some of the images. Did he look in a mirror or something, or did the video capture reverse it?
Since we are on the subject of Face Recolonization, when will some public use FR come into effect? I really like google's FR for online albums, but I have a hard drive with GIGS of pictures I would like to easily go through and find and name faces for me. I am hoping google or some company will release an offline FR software that we can use. Anybody know of any commercial software that does it?
get ready for a quad-core TiVo...
My favorite quote doesn't fit into 120 characters. Now no one will like me.
I suppose I should I be impressed that it cannot be fooled by beards, but I am instead frightened that the computer cannot tell the difference between real Spock and Evil Spock.
I can't wait until they start adding this feature to all those CCTV's...
You can try wearing a ski mask.
In fact, all this "privacy in public places" argument is kind of weird. The effect of these technologies is to bring the global village to a practical reality. You don't have too much privacy if you live in a small village. Everybody knows you were at Sally's house last night.
No, you got it wrong. Slashdot is a public place, but you aren't forced to walk naked in public places. Think of the Slashdot nick as digital clothes.
But still there's no real privacy here or anywhere else. Remember the Unabomber? He was caught because the sentences he used in his manifesto seemed familiar to his brother who hadn't seen or talked to him in ten years.
It seems perfectly possible that the police may use a software to analyze an AC posting in Slashdot and compare it to every signed blog in the web to find the true identity of the author.
Sadly, in the USA, copyright never dies. Ever. Just when you think it's about to, Disney pays off congress to extend it another 30 years. I expect by the time I die, copyrights will be lasting well over 250 years.
Your Welcome
Its not perfect. Barbara Babcock was recognized as 2 different faces, 00058 and 00061 and in the episode naked time, Majel Barret was recognized as Chapel and a random character, the only difference was hair up or down. Nice to know that a haircut can confuse it. I hope they don't start using something like this for profiling. Its not ready.
I reject your reality
Hmm... PittPatt claims they have an SDK that is usable by the public, but there is no download link. Does this seem suspicious to anyone else?
Most people don't get why the integral of "e to the x" is so funny. Most math majors don't have a sense of humor.
Click on the Spock faces at the top of Season Two (Amok Time, http://facemining.pittpatt.com/S2E30/) All Vulcan men look alike to this software.
Tivo already has a feature to show you actors in a show, and other shows they have been in. It will also tell you if the other shows they have been in have any hits on Amazon On-Demand you can stream to the Tivo or watch through YouTube on the Tivo.
I am a female lawyer you insensitive clod!
Are you also perchance... a SINGLE female lawyer?
You can't take the sky from me...
When a character rotates his head, so the camera gets the side, the recognition gets lost. The person then shows up as unknown. It sounds so simple to link a face that changes from known to unkown (rotating head) to each other, solving the unknown. Example is episode 66 at 6:53 where mccoy = unknown, when kirk moves in front of him, afterwards mccoy is known. There is a bit of a gap, but some reasonably smart algorithm should not be so difficult to cook up.
Well, they only analyzed TOS, so those examples wouldn't apply. However, it would pick up on Mark Lenard playing first a Romulan Commander before later portraying Sarak.
I think it would be interesting to identify the people in fight scenes. The HD conversion makes the stunt doubles even more obvious.
I wonder if it could match up with the faces in the animated series next.
Oh, say does that Star-Spangled Banner entwine / The myrtle of Venus with Bacchus's vine?