Insane? CUHK Claims 99.15% Recognition Accuracy

Department of Electronic Engineering, has built a novel facial recognition system that with the highest accuracy in the world. CUHK Research


Thoughts?

Login to read this IPVM discussion.
Why do I need to log in?
IPVM conducts unique testing and research funded by member's payments enabling us to offer the most independent, accurate and in-depth information.

**** ******* *******, **** ***** ****** *** ***** ** **:

"***** ****** ********* ***** ** ** ******** ****** ** **.**% on ******* ***** ** *** ****, *** *********** ****** ********* by **** *** ****** ***** ********* ** ******* ****, *** it *** ********* ***** ** ** ******** ** **.**%, ********** of ******* ** ********, ****-** *** ****** ******.

*****: "**.**%" ** *** **** ***** "*******" *****? *** ******* says '*********', *** ** ******* *** ** ******** ********* ** with *** ***** ****** **** ** '*********' ***** ** **** an ***** ******. ******** ******* ******** ** **** ****** **** large ***********. *****'* **** **** ** ******* *** ****** ********** among *********** ******* *****.

******: *****'* **** ** ***** **** ***** '******** **********', *** what ************ *** ***** ** *** ****** *****-***? ** *********** on ***, *********, ** ****** *********** ************ *** ******* - especially ** *** ***-******* *** **** ************ ******* *** ********.

*****: **** ***** '********' *** ** * ****, ** * single-file ***** **** ******** ** ***** ** * ******, ** was ** * **** ** ****** ******** ******** ** * crowd? * *** **** ***** *****'* ********* ** ********** ***.

** **, ******* **** ***** *** ** ********* ***** *** any ***** ********* **** *** *********. ** *** ***** ** impressive ******* **********, *** ** ********** ****, *** ******* ****** nothing **** **** ****** '****** *****' **** ** **** ******** technology ******** **** ** ********* ** ** ************* *********.

*** ****** **** ** *** ***** ****** ********** *************:

"** *** ********* ***** ** ** ******** ** **.**%,regardless ** ******* ** ********, ****-** *** ****** ******"

**'* ********** ********** *** ** *** ** ** ******** ** 'lighting *** ****** ******'.

***** ******* **** ****:

***** *** ***** *** *** ** **********...

****** **** ** *** **** ***** **** *** **************** *********** ***** *********...

*** ******** ********* ** ********* *** ********** (****) ******* **** results **** *** **** **** ** ****** *********** ********** **** that ******** *** ******** ** ** ** ******* ***** ****...

** **** ***** *** ***** **** * ******** ** *.* million *****. ** ****, *** ****** **** ******* "*******" ****** from *** *********** ******** **** ******* ****** **** *** ****/**** ITL *-**** **** ** ********. ** ****, *********** ***** * small ******** ** ****** ***** *** **** ************ **** **** an ***/*** (************* ************ *** ***************/************* **************** **********) ******** *** 140,000 ****** ****** ***** ** ****** ********** ************ **** ** not ****** **** *** ********...

*** ***** **** ***** **** ***** ** ******* ****** ******* increase ** *** ******** **** ********* ** ********, *** **** it **** ** **** ******. **** *** ****** ** ****** images ********* ** * ****** ** **—**** ***,*** ** *.* million—the ***** **** **** ********* ** ***** *.* *****. **** slower-then-expected ****** ** ***** ***** ****** ** **** ******* **********, and "** ******* *********** *** *** *********** ******* ** **** identification **********," ******** *******

*.*. * **** *** ********** **** ** *** ***** *** above:*** "*********" ******