Exacq "Facial Matching" Facial Recognition Tested
***** ** ******** "******** ****** ** Every ********** ****** ** ********" **** their**** ********* "****** ********", *** *** **** **** ** work ** ****-***** ******?
** ****** ***** ****** ******** *********:
- *** **** ** ******* ** ****-***** surveillance *********?
- *** ***** **** ** ******* ******** a ****** ** ******* ****?
- ***** **** ** **** ***** / people ******* ******* *** *****?
- *** ** ***** ********* ***** ** people ******* *****, **********, ****, *******?
- *** **** ** ******* ** ***** in *** ***** (~* ***) ** dark/IR (~*.** ***)?
- **** ** ****** ******** ***** **** human *****?
- ** *** **** ********* **** ** navigate?
*******
***** *****'* ****** ******** *** ** false ********* ** **** *** ***** of *******, ** *** ********* **** in ****-*** ****** (***+ ***) ** high ***** ******* (~*** *** ***********), with ******** ******* ******** ****** *** camera. ********** ******** ** ********** ****** past ~**-**°, ********** ***** ****** (**** and ******* ********** **), ** ******* masks *** ******** ** ********** ******, regardless ** ********** ********.
***** *********** *********** ***** ** ****** competitors **** ***********,********, *******, ***** ******* **** ****** *********** at ******* ******, ***** ****, *** lower *****. *******, ***** ****** ******** is ******* **** ********* (~$*** **** vs. ~$*,***, **** ** ******* ******* count).
************, ***** **** ******* *********** ** Exacq's **, **** ******** ***** *** video ******* *** ** ****, ******* setup ******** *** ****** ***** *************, and ** *** ** **** ****** subjects, *** ****.
***** ** ***** *******, ***** ****** Matching ** **** *********** *** *:* verification ** ***** ********, *.*. ****** / ***** *******, ***** ***** ******* for *:* **-************** ***** ** ****** served ** ******* ***** ****** **** challenging ************ ******.
Exacq ******** / ***** ***************
***** ********** ***** ***** ***************, *****, which ********* ******** ** ******** ** 200+ ***, ** *** ***, **** subjects ******** ****** *** ******.
********* ***** ***************, ****** ******** ********* well. *******, ***** ********* ** **** tested ** *** ******* **** **** PPF ****** (*********, ********, *** ******** all ********* ***** ******* **-*** ***), are **** ** ********* ** ******* angles, *** ******* **** *********** ** low ***** *** ** ******.
Relatively ***** ****** ****
** *** *****, ***** ****** ******** observed >*** ****** ********, ********* **** employees, *********, *** ******** ** ****** businesses, ** **** ** ******* ***** of ************ ******* ***** **** ******* sources ** *** ********.
*******, ***** ********** *** ********** ***** compared ** **** ************ **** *** is ****** ** ** ***********, **** as ******* **** ******* ******* ********, casinos **** ********* ** ********, *** city ******* **** **** ** ********* of ****** ********. ******* ** ****** populations *** ****, **** ***** *********** issues **********.
Accurate *********** ** **** *** ****** / ** ***** *********
****** *** ***** ** *******, ***** were ** ***** ************ ***** * database ** ~*** ****** ** *** around ** ******/********* *******.
Recognition **** ** ***** ********* ******* ******** ****** ******
*******, *********** *** **** ******** ** a **** ***** ******* (~*** ***) while ******* ******** ** *** ******.
* **** ***** *** ** ********** when *********** ** **° ** ****** or **** ********** ******** **** ~**°.
No *********** ** **** /** (~*.** ***) ** *** ***** (~* ***) ******
*******, *********** *** *** ******** ** a **** ***** (~*.** ***) **** IR **** ** ******* ********* (~**°) when ******* ******** ****** *** ******.
*********, *** ******** ****** ** ********* subjects ** * ~* *** *** light ***** ** ***** **** ** these ******.
Subject *** ********** ***** *******
****** *** *****, ** ***** **** one ******* *** ** ****** ********** after *******, ********* ** ** **-******** him ***** * *****-****** *****.
**** **** ** **** *** **** this ***** ** ***** ******* ** have ******, **** ** *********, ********, Herta, ***.
No ******** ********** ******* *****
****** ******* ***** ** *** ***** were *** **********, ********** ** ********** settings.
***** ********* **** ****** ****** *********** performance ***** ** ******** ** *********** subjects ***** ****** ******, *** ** found ****** ** ** ********** ** performance ** *** *****.
Faces ********** ** *** *****
** *** ******* ******** **** ********** recognized ** * *** ***** **** strong *********, **** ** *** ***** scene ***** **** * ******** ****.
People ********** ******* ********** / ****
******** ******* ******* *** ***** ******* sunglasses *** **** **** ********** *** only ** ******* ********* ** ~**°.
*********, ******** ******* **** **** ********** only ** ***** ******** ******* ** ~7' *** ********* ** ~**°.
* *********** ** **** ******** ** missed ************.
******* ** *****, ***** ********* **** improvements ** *********** **** ******* ***** be **** ** *********** ******** ***** wearing *******, *** ** ***** ******* improvement ** *** *******.
Unsynchronized ******** *****
** *** *******, ******** ***** **** frequently *** ************ **** ****** *******, resulting ** *** *** ******* ****** the ******* (*****, **** ******) ** appearing ** ***** ** ****.
**** ******** ********** ** *************. ***** server, ** ******, *** ****** ***** were *** *** ** *** **** time ******, ** *********** ** *****.
Server ************ / *****
*** **** ** ****** ** ******** as ** ****** ************ *** ******** a ********* ******. **** *********, *** server ** ***** ** *********** ** a ******:
**** *****, ******* *** ***** *** configured *** * *** ****, ****** titled "** *******":
Only *** ****** ** * **** ***** ** ********
****** ******** ** *** ******** ** simple, *** *** ** **** ***** still ****** ** **** ***** ******. However, ***** ** ** ****** *** bulk ******** *********, **** ****** ********.
***** **** **** **** **** ****** with **** ********* ** ****** ****** databases, *** ** ** *** * standard ******* ** *** ***** **. They *** *** ** ** ****** releases ********* ** ******** ******.
*******
**** ****** *********** ***** *** $*** per ****** *******, **** ** ******* channel *****. *********** ********** ** ******** for ****** ******** *********** (~$***/*******).
**** ******* ** ************* ***** **** some ****** *********** *********, **** ** Anyvision ** ***** (~$*,***-*,***/*******), *** **** not **** *** *** ******* ******* of ********, ** ***** ** ******* to * ****** ******.
Versions ****
*** ********* ******** **** **** ****** testing.
- **** **:*.*.*.***
- *********** **********:**.**.*.*