Strut Your Stuff: Biometric Gait Recognition is a Reality

[presence]

Serious privacy concerns have been aired about the system and its potential surveillance applications.

http://rt.com/news/identify-walk-system-britain-668/

A new biometric “gait recognition” system has been developed by Britain’s National Physical Laboratory, meaning that individuals can now be recognized and located by their “signature” walk.

*

New Scientist reports that NPL, which collaborated with the Center for Advanced Software Technology (CAST), the BBC and BAE Systems, developed a new system through which a person’s walk could be identified. The tracking system combines a computer model of the NPL building with feeds from each on-site CCTV camera.

In each video frame, the system separates an individual’s silhouette from its background. The rise and fall of head height is recorded, and the pattern it forms can be represented by a set of numbers. This is linked to the person’s identity.

The work formed part of a larger project, called VSAR (Viewers Situational and Spatial Awareness for Applied Risk and Reasoning)

http://www.npl.co.uk/news/walk-this-way

I'm telling you... its coming: Every person isolated, numbered, and tracked on one central computer constantly by satellite and drone. A real time map of little red dots pulsating with data to represent each of us, moving around like tagged wildlife with our life long trails and tales remembered forever.



presence

 

[MelissaWV]

That has got to be the dumbest and flimsiest method of recognition I could possibly think of.

My back hurts today; I am not walking normally. If I've been walking a long time, there is definitely less spring in my step. If it's cold, I will almost jog to get to where I'm going. If there is ice I may change my stance and gait to help me navigate slippery conditions. If I am in heels I will walk differently than when I am in flats.

 

[CaseyJones]

 

[presence]

That has got to be the dumbest and flimsiest method of recognition I could possibly think of.

My back hurts today; I am not walking normally. If I've been walking a long time, there is definitely less spring in my step. If it's cold, I will almost jog to get to where I'm going. If there is ice I may change my stance and gait to help me navigate slippery conditions. If I am in heels I will walk differently than when I am in flats.

Does the computer lose track of the hurricane when the wind speed changes or the eye wall collapses? In all those instances you still have a last known location, trajectory, and biometric bone structure. Heels or no heels, walk, jog, or run... the computer knows its you; especially when it already tracking and has ruled out everyone else around you.

 

[TheTexan]

I too gotta call BS on this. This definitely isn't possible yet. That's not saying they wouldn't implement it anyway though. Shitloads of false positives? I don't think that's really a concern of theirs.

 

[presence]

I too gotta call BS on this. This definitely isn't possible yet.

I have a friend that programs for darpa and the national weather center, from the small collection of compartmentalized projects I've discussed with him I can assure you: This isn't BS and definitely is within the realm of possible. Read through the NPL link in the OP.

 

[TheTexan]

I have a friend that programs for darpa and the national weather center, from the small collection of compartmentalized projects I've discussed with him I can assure you: This isn't BS and definitely is within the realm of possible. Read through the NPL link the OP.

On a small scale, sure. On a larger scale, when you're monitoring millions of people, it simply wouldn't be possible without 200,000 teraflops of processing power...


oh wait..

shit

 

[presence]

200,000 teraflops of processing power...


shit

You're seeing the light

 

[GeorgiaAvenger]

 

[TheTexan]

Looks like they are measuring 3 metrics to create the signature, 1) distance between cycle start/end, 2) height, 3) stride length

However height has a strong influence on stride length, which has a very strong influence on the distance between cycle start/end.

IMO you would need to collect more data than just that to be able to identify an individual among tens of thousands, let alone millions.

 

[presence]

Looks like they are measuring 3 metrics to create the signature, 1) distance between cycle start/end, 2) height, 3) stride length

Cycle, height, and length: Thats tempo, volume, and melody if you think about it. Pretty unique song / set of metrics.

 

[acptulsa]

And you thought that was just silly.

Monty Python yet again proves forty years ahead of its time.

 

[TheTexan]

Cycle, height, and length: Thats tempo, volume, and melody if you think about it. Pretty unique song / set of metrics.

Except cycle, height, and stride length are all heavily related to each other. So your analogy doesn't really fit.

 

[presence]

I think some of you guys are missing the way this works.

If you take a known gait from an individual in 1995, the computer may or may not be able to match it to the same person in 2012, without continuous game play from 1995-2012. But if you give the computer a game to play: Isolate and track the ten people in this room. With accurate sensors and robust software, the computer will be able to assign each a number and track them indefinitely as individual entities providing you with a realtime map of their respective locations with an endless trail of where they were and when. Can the computer track 100 individual gaits simultaneously? The PC I'm typing on can probably do that with the right software and sensors. 10,000 in the stadium? The local university campus can handle it. 100,000 at this protest? A million? Probably.

6.5 billion on this large sphere?

Do any of you understand what a petaflop is? I've tried to understand and I still don't. But I expect along with gait data, last known facial recognition data, last known credit card purchase, cell phone gps, et al... there is a very high pitched hum in a very cold room deep underground that can handle it.

 

[MelissaWV]

Does the computer lose track of the hurricane when the wind speed changes or the eye wall collapses? In all those instances you still have a last known location, trajectory, and biometric bone structure. Heels or no heels, walk, jog, or run... the computer knows its you; especially when it already tracking and has ruled out everyone else around you.

Not sure whether or not you think you are addressing my post.

That has got to be the dumbest and flimsiest method of recognition I could possibly think of.

Nothing that you have said counters the fact it is very flimsy. My stride length today was entirely different from most days. There are times I have been known to be on crutches. There are times I walk with a cane. I have, on occasion, been in a wheelchair. If I knew this system was in place I could change my "gait" by doing any of those things, or are you contending that it will measure my wheeling around from a last known location in a giant crowd of people?

This will not hold up in court, and I doubt the military would even get much out of this over time. As a reinforcing means of identification, it could have some value, but the idea of tracking one individual who knows they are being tracked this way? Not so much.

 

[acptulsa]

First time I ever thought of corns and/or bunyons as something that could be desirable under certain circumstances.

 

[TheTexan]

I think some of you guys are missing the way this works.

If you take a known gait from an individual in 1995, the computer may or may not be able to match it to the same person in 2012, without continuous game play from 1995-2012. But if you give the computer a game to play: Isolate and track the ten people in this room. The computer will be able to assign each a number and track them indefinitely as individual entities providing you with a realtime map of their respective locations with an endless trail of where they were and when. Can the computer track 100 individual gaits simultaneously? The PC I'm typing on can probably do that with the right software and sensors. 10,000 in the room? The local university campus can handle it. 100,000? A million? Probably.

Tracking gaits is very different from using gaits as an identifier though. The technology to track a moving object has been around since the stone age. Using it as an identifier is an entirely different story.

To use it as an identifier, without false positives, it obviously follows that every individual must have a unique gait. However I do not believe that is true. There's only a limited amount of variations... and making it more complicated, many people will vary their gait based on a number of factors (if they are in a rush, if they are trying to look cool, if they are nervous, etc).

To get the level of detail required would probably require 50 NSA secret spy centers all working at full capacity on this single task... when it would be so much easier just to use facial recognition

 

[presence]

Except cycle, height, and stride length are all heavily related to each other. So your analogy doesn't really fit.

Instead of thinking of the 3 variables as numbers, think of them as stocks and picture a historic chart for each. The three metrics become a historical fingerprint with known cycles and patterns. Each individual may not have a unique stride length at the moment, but each has its own unique historical graph with unique patterns.

 

[pcosmar]

Walk without rhythm, and it won't attract the worm ...

 

[tod evans]

If we're reading about it, it's been in play for a while.