Biometrics research considers quite a multitude of human (and animal) body properties, traditional - fingerprint, iris, face, hand geometry, and not so traditional - palmprint, ear geometry, gait, blood vessels, chemical composition, teeth, etc. The popularity of biometrics is mostly determined by the convenience of its usage. Clearly, genitalias and anuses will provide great biometrics, but their usage is very limited. Described application makes perfect sense. No more time to waste, have to write a new research proposal!
I read long time ago somewhere that 1.5 (previously intended to be 2.0) is supposed to fix memory bugs. Unfortunately, I could not find any statement from description page that memory leaks were indeed fixed in this version.
Anyway, here is a bug: https://bugzilla.mozilla.org/show_bug.cgi?id=24946 9 (copy and paste in new tab). Note how many bugs were marked as duplicates of this bug. Do we need to submit more bug reports on this bug, or will somebody working on it clarify the situation?
Some extension was specifically created to save session so that people would be able to save session, exit FF, free leaked memory, start FF, and load saved session. This is really ridiculous - to have some extension to bypass bug instead of fixing bug.
In my case I have 512 Mb of memory and have to restart FF at least once a week. This is quite undesirable for my work since I like to keep some pages open for reference (scientific research) for a while. Also, few sites require login, and restarting FF even with that extension is not that simple.
In my opinion, this bug is the most important thing that makes me question the usability of FF. Really, I need to adjust my browsing behavior in order to deal with this bug, and constantly check task manager (>300Mb => restart).
I admit I do not have access to IBG test reports. If you have access, good for you. The numbers which you cite probably come from Bio-key but I tend to be cautious about them. Anyway, it is a question whether you believe some company's propaganda or not. Presenting better numbers directly translates into bigger sales and company's livelihood. Tweaking numbers is unfortunately a frequent way of doing business (I am not asserting this happened here; I simply do not have enough information to judge).
Testing setup usually greatly influences performance results. "Cleaning", that is removing few (5-10) fingerprints from FVC2002 database would make most fingerprint systems perform perfectly. There is no information how bio-key's algorithm was tested. It would be really interesting to see where this 1:200,000,000 figure comes from.
New crime type still works: scan your unenrolled middle finger with random phone number.
You really have to look beyond PR bullshit and see if technology worth something. It is rather hard to do with bio-key, only information being rather unreadable patent.
I have some reservations about biometric systems since I know how most of them work. "Population of millions" is still remote future for biometrics. You need accuracy, you need speed (indexing), you need some standard protocols. The key to solving many problems is combination. There are scanners which can determine liveness of a skin and there are fingerprint scanners. There should be: combined fingerprint and liveness scanner. ETA: 2-3 years. Almost perfect scanner: all-in-one fingerprint, liveness, chemical, blood vessel, hand geometry, palmprint, (DNA?). ETA: 10-20 years.
The main critique of the proposed system: the protocol is not standard. The fingerprint should really be combined with some token, like credit card. The required phone number (or any other number) is really the key for this system but company's sales department is at work here, and it is conveniently forgotten when vaunting the product.
New crime type (completely innocent): I pay for groceries and 'accidentaly' enter wrong phone number. Scan finger, if does not work enter correct phone number and proceed. After 100-1000 shopping trips and changing 'accidental' phone numbers: lo and behold - some phone number (not mine) worked with my fingerprint! Forget my own phone number and go to a shopping spree!
Slashdot Mirror
Biometrics research considers quite a multitude of human (and animal) body properties, traditional - fingerprint, iris, face, hand geometry, and not so traditional - palmprint, ear geometry, gait, blood vessels, chemical composition, teeth, etc. The popularity of biometrics is mostly determined by the convenience of its usage. Clearly, genitalias and anuses will provide great biometrics, but their usage is very limited. Described application makes perfect sense. No more time to waste, have to write a new research proposal!
Some extension was specifically created to save session so that people would be able to save session, exit FF, free leaked memory, start FF, and load saved session. This is really ridiculous - to have some extension to bypass bug instead of fixing bug.
In my case I have 512 Mb of memory and have to restart FF at least once a week. This is quite undesirable for my work since I like to keep some pages open for reference (scientific research) for a while. Also, few sites require login, and restarting FF even with that extension is not that simple.
In my opinion, this bug is the most important thing that makes me question the usability of FF. Really, I need to adjust my browsing behavior in order to deal with this bug, and constantly check task manager (>300Mb => restart).
Testing setup usually greatly influences performance results. "Cleaning", that is removing few (5-10) fingerprints from FVC2002 database would make most fingerprint systems perform perfectly. There is no information how bio-key's algorithm was tested. It would be really interesting to see where this 1:200,000,000 figure comes from.
New crime type still works: scan your unenrolled middle finger with random phone number.
You really have to look beyond PR bullshit and see if technology worth something. It is rather hard to do with bio-key, only information being rather unreadable patent.
I have some reservations about biometric systems since I know how most of them work. "Population of millions" is still remote future for biometrics. You need accuracy, you need speed (indexing), you need some standard protocols.
The key to solving many problems is combination. There are scanners which can determine liveness of a skin and there are fingerprint scanners. There should be: combined fingerprint and liveness scanner. ETA: 2-3 years. Almost perfect scanner: all-in-one fingerprint, liveness, chemical, blood vessel, hand geometry, palmprint, (DNA?). ETA: 10-20 years.
The main critique of the proposed system: the protocol is not standard. The fingerprint should really be combined with some token, like credit card. The required phone number (or any other number) is really the key for this system but company's sales department is at work here, and it is conveniently forgotten when vaunting the product.
New crime type (completely innocent): I pay for groceries and 'accidentaly' enter wrong phone number. Scan finger, if does not work enter correct phone number and proceed. After 100-1000 shopping trips and changing 'accidental' phone numbers: lo and behold - some phone number (not mine) worked with my fingerprint! Forget my own phone number and go to a shopping spree!