Active Appearance Models work by creating a deformable model of appearance built by combining a point distribution model with a texture model using principal component analysis (PCA). Basically what that means you take a bunch of faces, located landmark points like the corners of eyes, apex of the chin, etc., create an average face and use PCA to statistically model the variations. Next you morph the faces together from their original landmark points to average face, and do a PCA on the pixel values. This creates a pixel-wise model of 'texture' which also models variations. With these two parts you have a thing that (with a good sampling of face) models most faces and emotions with about say 80 to 120 numbers and statistical ranges for those numbers.
So how do you use this to track faces? Well you use gradient decent to optimize the appearance of the face to image by adjusting those 80-120 values, x, y, scale, rotation until the pixel difference is close to zero. The trick is the gradients are approximated by precomputed derivative images, but this only works if the model is initialized on top of the original face. You can see in the video, he used Viola-Jones (the green squares) to locate the face and then dropped the AAM on top of it. He's only showing the landmark points and not the texture model.
I did my dissertation on this almost a decade ago for tracking MRs of hearts, even back then it was pretty fast. What's interesting is not only can the model identify, but you can also reconstruct synthetic images of faces, and the model parameters could be used for identifying a person, identifying an emotion, creating a synthetic face swapping another person's identity but keeping the same parameters for expression, etc. My own implementation reliably detected anomalies in beating hearts.
I really wanted to build a business around it back then, but it was in conflict with my advisor and university at the time.
My mistake...I just saw the video where Hinton was suggesting to Google, tongue-in-cheek, to use his RBM bottleneck trick as a method of databasing and then to see this guy's paper mentioning the very same thing a few weeks later.
It was a skim to see what the hell the article was really about, didn't know these two were connected. I jumped the gun 'cause I got burned by a plagiarizer in the past, sorry.
I hate reading press releases of reading papers with real explanations of what's going on.
I just finished reading "Small Codes and Large Image Databases for Recognition" written by the guy. All he did was implemented Geoff Hinton's idea of databasing images with a binarized coefficients produced by Restricted Boltzmann Machines.
total = (500,000 x 10^6 g of shit) * (1.4 x 10^-3 liters of gas) / (100 g of shit) total = 7,000,000 liters of gasoline total = 1,849,204.36 US gallons
4000 people would benefit. Woohoo!!! Energy Independence! USA! USA! USA! USA!
We're not even factoring the gasoline needed to make the process work or to ship the shit/gasoline in and out of the plant. The problem with most policies makers even business people is that they don't do "back of the envelope calculations" before passing laws or making decision, especially with science related stuff.
One problem, even if it were feasible to clone a T-Rex (which mostly likely it isn't) there is the tiny fact that dinosaurs at the time lived in a higher oxygenated atmosphere. This made it possible for them to grow as large as they did.
It seems like a logical progression. I remember when Slashdot was "Chip'n Dip", and people mostly talked about computers and gadgets. Then as we approached the Great Internet Bubble, business related topics seemed to take over. Now that we've moved into post-Bubble recovery, focus has shifted to legal wranglings created by the after effects of the bubble or people desperately trying to make a quick buck like others once did.
What's next? Articles about surviving the post-dollar crash depression? "Cob/Mud built houses aren't that bad after all.", "Welcoming our Chinese Overlords.", "Programming for Food? HTML for Handouts."
So imagine a scenario when globalization collapses and no one has the ability or technology to manufacture complex goods like computers. All we'll have left is the remnants of circuit boards that will either be used to hobble together computing devices or art because of its increasing rarity.
My mom's side of the family, I being half-Korean, does a good job of maintaining the Korean Dry Cleaning stereotype, so I have a good insight on how dry cleaning works.
This type of device has existed since I can remember. Basically, you place the shirt/blouse/jacket onto a manikin-like device. A handle is pulled to mechanically stretch the item, and a lengthy blast of steam billows out until the wrinkles are gone. Nothing new here, except I guess the device is for home use and you do not need an external source of steam.
Maybe that's the trick: 1. Take something that exists in the professional field. 2. Re-design it for home use and call it a robot. 3. Profit.
No mysterious ???? step. I think we got something here.
What's going on in this country? Why are people taking extremely polar-opposite views (liberal vs. neoconservatism) attacking the other like little children throwing tantrums? Geez, it's gotten really bad these past fewer months, not to mention witnessing a real fistfight over politics. If you try stating facts either way you're chastised as a liar or a traitor. Just take a look at the previous posts above, where are the facts? Are we heading towards a left/right civil war?
He's a little kooky with his free-energy and UFO stuff, but you have to give him a hand with all the various experiments, videos, and analysis, designes, etc. that he's accomplished. I also love his Frenchy "This iz Jean Louis Naudin. I now turn ON zee lifterrr" in his real videos too.
A few friends of mine and I managed to build a working lifter using the usual balsa, foil, wire and an old monitor. We talked to a Physics Prof, and his conclusion was the best test to determine if it's ionic wind that's propeling the device or not (since vacuum will cause breakdown) is to test it inside a container of helium. The order of magnitude less of atomic weight should significantly reduce any effects of ionic wind, hence test if that's what causes it to fly.
If you think that is crony capitalism, check out Donald Rumsfeld and Aspartime (Nutrasweet). He was CEO and chairman of G.D. Searle & Co during 1977 to 1985 to add some political pressure on the FDA to approve Nutrasweet after Searl got caught falsifying the data about a bunch of monkeys and rats dying.
His Bio: http://www.defenselink.mil/bios/rumsfeld.html The nutrasweet controversy (if you believe them): http://www.swankin-turner.com/hist.html h ttp://www.rense.com/general33/legal.htm http://pr esidiotex.com/nutrapoison/
What totally drives me nuts is most people in that field are totally hooked on the whole fisher-face, eigen-face, ICA, thing. Basically they naively project a two-dimensional affine/brightness normalized face onto a basis function and then do a nearer neighbor on the coefficients to determine identity using some magical distance metric like Mahalanobis or Euclidian. They totally fail when the intensity or pose changes, and then blame it on the distance function or basis function.
Shape models and combined models take this into account and are really popular in medical imaging, yet the facial people seem to shoot down. (Well it's antidotal on my part).
Sorry, I guess I'm geeking out, but I love this stuff.
Most of my research involves adapting the `what's in' in facial recognition and applying it to disease diagnosis and segmentation of heart MRIs. These days' people at developing statistical shape and appearance models of faces via PCA for matching and segmentation. It gets a bit scary what they can do in facial recognition if you start reading up, but it's also slightly disconcerting how much money is being tossed in medical imaging.
Wrong, H3 cannot be made into a weapon because the amount of energy it takes (a nuclear explosion) to make it explode is greater than the resultant energy created. (Endothermic). Pretty useless for a bomb, huh?
According to a lecture on this subject by Dr. Gerald L. Kulcinski (University of Wisconsin-Madison) a working He-3 + He-3 fusion would change everything. 2He-3 ---> He-4 and 2 Protons!!
Just collect the protons and you got pure electricity, a nearly 100% conversion. No neutrons = no radiation! Hell you could have one running in the basement in theory. 40 Tonnes of He-3 (one cargo load on a space shuttle...like it could haul stuff from the moon) could produce enough power the entire US for one year.
I hate Bezier splines with a passion because they do not make intuitive sense. It makes more sense to use Catmull-Rom Splines where the control points fall on the line. It uses a simple parametric polynomial just like the Bezier, but unfortunately it lacks the subdivided-ness.
According to my Russian wife, she claims most hospitals in Russia control Staph and other nasties using UV lamps placed in hallways and rooms. If the room or hallway is vacated and the doors are shut, the lamps automatically go on killing most of the bacteria without the risk of developing resistance. Other than potential risk of exposure which probably could be kept under control, I think it's a great idea and I'm wondering why aren't US hospitals using it.
Maybe it's just me, but my biggest complaint is most computer books out there are concentrating on how to use the newest coolest language instead of the underlying principles. I'd rather have pseudo-code cover how to pattern a peer-to-peer network, an mp3 codec, a nearest word match spell checker, a regular expression engine, or a typical Civ-like AI. These days I hunger for books to explain how the hell Divx works without trudging me with specifics like how to fashion an if statement in Java or an STL in C++. I want material with reasonable amounts of math and code snippets, not a rehashed programming lesson.
I wish Unix/Linux had a mechanism where a directory could be marked executable and executing the directory whould internally call some default dot file (such as.name_of_directory)within the directory, and some environmental variable (like $THIS_PATH) was set to the directory and passed to the application process.
Maintance for applications like these whould be a no-brainer. Just move the directory and all the associated preference files and whatnot travel with the app.
So has anyone thought of an open source variant on Replay/Tivo? Consider you can get ATI video cards that can capture video in real time and hard drives for nothing, I could see maybe a specialized linux project to create the ultimate home converge box. Perhaps even a specialized linux distribution that could turn that spare high-end PC into headless super box beyond Replay and Tivo. Think of the cool projects that could revolve around this? Open Source databases of program listings that wouldn't be subscription based, IR interface for a remote, voice mail which would screen out telemarketers and display phone numbers on the screen, DVD, voice command channel changing, dim the lights, mix drinks, etc.
Slashdot Mirror
Active Appearance Models work by creating a deformable model of appearance built by combining a point distribution model with a texture model using principal component analysis (PCA). Basically what that means you take a bunch of faces, located landmark points like the corners of eyes, apex of the chin, etc., create an average face and use PCA to statistically model the variations. Next you morph the faces together from their original landmark points to average face, and do a PCA on the pixel values. This creates a pixel-wise model of 'texture' which also models variations. With these two parts you have a thing that (with a good sampling of face) models most faces and emotions with about say 80 to 120 numbers and statistical ranges for those numbers.
So how do you use this to track faces? Well you use gradient decent to optimize the appearance of the face to image by adjusting those 80-120 values, x, y, scale, rotation until the pixel difference is close to zero. The trick is the gradients are approximated by precomputed derivative images, but this only works if the model is initialized on top of the original face. You can see in the video, he used Viola-Jones (the green squares) to locate the face and then dropped the AAM on top of it. He's only showing the landmark points and not the texture model.
I did my dissertation on this almost a decade ago for tracking MRs of hearts, even back then it was pretty fast. What's interesting is not only can the model identify, but you can also reconstruct synthetic images of faces, and the model parameters could be used for identifying a person, identifying an emotion, creating a synthetic face swapping another person's identity but keeping the same parameters for expression, etc. My own implementation reliably detected anomalies in beating hearts.
I really wanted to build a business around it back then, but it was in conflict with my advisor and university at the time.
My mistake...I just saw the video where Hinton was suggesting to Google, tongue-in-cheek, to use his RBM bottleneck trick as a method of databasing and then to see this guy's paper mentioning the very same thing a few weeks later.
It was a skim to see what the hell the article was really about, didn't know these two were connected. I jumped the gun 'cause I got burned by a plagiarizer in the past, sorry.
I hate reading press releases of reading papers with real explanations of what's going on.
I just finished reading "Small Codes and Large Image Databases for Recognition" written by the guy. All he did was implemented Geoff Hinton's idea of databasing images with a binarized coefficients produced by Restricted Boltzmann Machines.
Hinton himself gave a talk on it for Google here:
http://www.youtube.com/watch?v=AyzOUbkUf3M
Actually I'm wondering, is he plagiarizing Hinton?
Damn it, who's going to buy my captcha solvers now?
total = (500,000 x 10^6 g of shit) * (1.4 x 10^-3 liters of gas) / (100 g of shit)
o line_per_capita.html)
total = 7,000,000 liters of gasoline
total = 1,849,204.36 US gallons
Per capita consumption 464 gallons per year
(http://www.energy.ca.gov/gasoline/statistics/gas
Americans = (1,849,204.36 US gallons) / (464 gallons)
Americans = 3,985.3
4000 people would benefit. Woohoo!!! Energy Independence! USA! USA! USA! USA!
We're not even factoring the gasoline needed to make the process work or to ship the shit/gasoline in and out of the plant. The problem with most policies makers even business people is that they don't do "back of the envelope calculations" before passing laws or making decision, especially with science related stuff.
One problem, even if it were feasible to clone a T-Rex (which mostly likely it isn't) there is the tiny fact that dinosaurs at the time lived in a higher oxygenated atmosphere. This made it possible for them to grow as large as they did.
-Steve
It seems like a logical progression. I remember when Slashdot was "Chip'n Dip", and people mostly talked about computers and gadgets. Then as we approached the Great Internet Bubble, business related topics seemed to take over. Now that we've moved into post-Bubble recovery, focus has shifted to legal wranglings created by the after effects of the bubble or people desperately trying to make a quick buck like others once did.
What's next? Articles about surviving the post-dollar crash depression? "Cob/Mud built houses aren't that bad after all.", "Welcoming our Chinese Overlords.", "Programming for Food? HTML for Handouts."
-Steve
So imagine a scenario when globalization collapses and no one has the ability or technology to manufacture complex goods like computers. All we'll have left is the remnants of circuit boards that will either be used to hobble together computing devices or art because of its increasing rarity.
Actually an inch is defined to be exactly 2.54 cm. Therefore you need the meter in order to standardize US length measurements.
My mom's side of the family, I being half-Korean, does a good job of maintaining the Korean Dry Cleaning stereotype, so I have a good insight on how dry cleaning works.
This type of device has existed since I can remember. Basically, you place the shirt/blouse/jacket onto a manikin-like device. A handle is pulled to mechanically stretch the item, and a lengthy blast of steam billows out until the wrinkles are gone. Nothing new here, except I guess the device is for home use and you do not need an external source of steam.
Maybe that's the trick:
1. Take something that exists in the professional field.
2. Re-design it for home use and call it a robot.
3. Profit.
No mysterious ???? step. I think we got something here.
*Sigh* It's crap like that, which gives my alma mater a bad name. As a fellow Hawkeye, I apologize for this.
:(
I should have gone to ISU or UW Madison. Oh the regrets.
"Damn bugs wacked us Johnny!"
The weird part is, the more I watch that movie, the more I like.
What's going on in this country? Why are people taking extremely polar-opposite views (liberal vs. neoconservatism) attacking the other like little children throwing tantrums? Geez, it's gotten really bad these past fewer months, not to mention witnessing a real fistfight over politics. If you try stating facts either way you're chastised as a liar or a traitor. Just take a look at the previous posts above, where are the facts? Are we heading towards a left/right civil war?
Oh come on now, THE best site to go to for this stuff is:
http://jnaudin.free.fr/lifters/main.htm
He's a little kooky with his free-energy and UFO stuff, but you have to give him a hand with all the various experiments, videos, and analysis, designes, etc. that he's accomplished. I also love his Frenchy "This iz Jean Louis Naudin. I now turn ON zee lifterrr" in his real videos too.
A few friends of mine and I managed to build a working lifter using the usual balsa, foil, wire and an old monitor. We talked to a Physics Prof, and his conclusion was the best test to determine if it's ionic wind that's propeling the device or not (since vacuum will cause breakdown) is to test it inside a container of helium. The order of magnitude less of atomic weight should significantly reduce any effects of ionic wind, hence test if that's what causes it to fly.
If you think that is crony capitalism, check out Donald Rumsfeld and Aspartime (Nutrasweet). He was CEO and chairman of G.D. Searle & Co during 1977 to 1985 to add some political pressure on the FDA to approve Nutrasweet after Searl got caught falsifying the data about a bunch of monkeys and rats dying.
e nutrasweet controversy (if you believe them):
h ttp://www.rense.com/general33/legal.htmr esidiotex.com/nutrapoison/
His Bio: http://www.defenselink.mil/bios/rumsfeld.html
Th
http://www.swankin-turner.com/hist.html
http://p
etc...just Google Donald Rumsfeld and Aspartime.
What totally drives me nuts is most people in that field are totally hooked on the whole fisher-face, eigen-face, ICA, thing. Basically they naively project a two-dimensional affine/brightness normalized face onto a basis function and then do a nearer neighbor on the coefficients to determine identity using some magical distance metric like Mahalanobis or Euclidian. They totally fail when the intensity or pose changes, and then blame it on the distance function or basis function.
Shape models and combined models take this into account and are really popular in medical imaging, yet the facial people seem to shoot down. (Well it's antidotal on my part).
Sorry, I guess I'm geeking out, but I love this stuff.
Most of my research involves adapting the `what's in' in facial recognition and applying it to disease diagnosis and segmentation of heart MRIs. These days' people at developing statistical shape and appearance models of faces via PCA for matching and segmentation. It gets a bit scary what they can do in facial recognition if you start reading up, but it's also slightly disconcerting how much money is being tossed in medical imaging.
According to a lecture on this subject by Dr. Gerald L. Kulcinski (University of Wisconsin-Madison) a working He-3 + He-3 fusion would change everything. 2He-3 ---> He-4 and 2 Protons!!
Just collect the protons and you got pure electricity, a nearly 100% conversion. No neutrons = no radiation! Hell you could have one running in the basement in theory. 40 Tonnes of He-3 (one cargo load on a space shuttle...like it could haul stuff from the moon) could produce enough power the entire US for one year.
Check out these slides.
I hate Bezier splines with a passion because they do not make intuitive sense. It makes more sense to use Catmull-Rom Splines where the control points fall on the line. It uses a simple parametric polynomial just like the Bezier, but unfortunately it lacks the subdivided-ness.
Look into it.
According to my Russian wife, she claims most hospitals in Russia control Staph and other nasties using UV lamps placed in hallways and rooms. If the room or hallway is vacated and the doors are shut, the lamps automatically go on killing most of the bacteria without the risk of developing resistance. Other than potential risk of exposure which probably could be kept under control, I think it's a great idea and I'm wondering why aren't US hospitals using it.
-Steve
Here's a great patent: Method for publicly securing exclusive control and possession of an invention.
Some one needs to patent this concept and sue the crap out of the patent office for infringement.
-Steve
So, just being curious...how many people here actually found Spyware using "WHO'S WATCHING ME?"?
Not me.
Maybe it's just me, but my biggest complaint is most computer books out there are concentrating on how to use the newest coolest language instead of the underlying principles. I'd rather have pseudo-code cover how to pattern a peer-to-peer network, an mp3 codec, a nearest word match spell checker, a regular expression engine, or a typical Civ-like AI. These days I hunger for books to explain how the hell Divx works without trudging me with specifics like how to fashion an if statement in Java or an STL in C++. I want material with reasonable amounts of math and code snippets, not a rehashed programming lesson.
One of these days I'll write that encyclopedia.
I wish Unix/Linux had a mechanism where a directory could be marked executable and executing the directory whould internally call some default dot file (such as .name_of_directory)within the directory, and some environmental variable (like $THIS_PATH) was set to the directory and passed to the application process.
Maintance for applications like these whould be a no-brainer. Just move the directory and all the associated preference files and whatnot travel with the app.
-Steve
So has anyone thought of an open source variant on Replay/Tivo? Consider you can get ATI video cards that can capture video in real time and hard drives for nothing, I could see maybe a specialized linux project to create the ultimate home converge box. Perhaps even a specialized linux distribution that could turn that spare high-end PC into headless super box beyond Replay and Tivo. Think of the cool projects that could revolve around this? Open Source databases of program listings that wouldn't be subscription based, IR interface for a remote, voice mail which would screen out telemarketers and display phone numbers on the screen, DVD, voice command channel changing, dim the lights, mix drinks, etc.