China Facial Recognition Jaywalking Market 2021
In 2018, China's facial recognition systems for identifying jaywalkers became global news. But what has happened in the few years since?
In this report, we examine the 2021 market for China facial recognition jaywalking systems, including:
- Examining the process of punishing jaywalkers including automated and manual actions
- Sharing feedback from Dahua, Tiandy, and Uniview engineers on these systems' challenges
- Detailing the providers of these systems including Hikvision
- Estimating how broadly these systems are being used and their growth rate
- Explaining how punishment works for jaywalkers
- Elaborating on the appeals process
- Habit building as a key purpose of systems
- How local governments ppost jaywalkers on social media
- What netizens concerns are
For our original coverage on these systems, see IPVM's "China Jaywalking Facial Recognition Guide".
Executive *******
******** ****** *********** ********** ******* **** received *********** *********, ***** ********* ********* are ***** ***********. ******* ***** ********* use ****** ************. *******, ************** ******* limited ****** ***** *+ ***** ***** as **** * *** ** *** busiest ************* *** *******. ***** ****** are ********* *** *** ** ***** systems, ******** **** ***** **** ***** supposed *************.
*******, **** ** ***** ******* **** limitations, *** ********* **** ** ***** and ****** ****** ***** ****** *** help ****** **********.
**********
****** *********** ************ ********* ** ***** ** ****** ***** **********, * ****-******** *******. These ******* *** *** ** ********** crossings *** **** ******** ** ********** as **** ***** *** ****. *** system ***** ******** * ***** *** screen, **** ** *** *** *****, that **** **** **** ******** ** the ********* ******** *** **** ********* as **** ** * ****** ******* of *** *********'* ****.
*******, ** **** ******** ** ****, the "***** ** *****" ** *** *** **** *** of ********* **********, ** ***** **** been ******* ** ********** ** ********, **** ********, ******* ***** **** *** ***** ****** **********.
Little **** ** **** **********
******** **** **** ******* ******* ***** facial *********** ******* ************* *******, ********, and **** **********, **** ********** ** generally *** ***** ****. ***** ******* typically ***** ****** ** ****** ** manually ******* ******* ****** ******* *** fines.
*****'******** ****** *** *** ********** ****** ***** a ********** ** *** ******, **** a ****** (***** ** *** *****) labeled "******* ********" (审核确认), ***** * human ***** ******** **** ** ***** to ******* *** ***** ***** ** the ******.
****** **** ****** ******* ** *** Fuzhou ********** ******** ****:
*** **** *********** ****** ************* ******** the ******** *********** ** *** ******** and ***** ** ** *** *** enforcement ******** ***manual ****** *** ************. [emphasis added]
*** ******* ******** ** ***** ******* how *** ****** ***** ******** **** on *** *******. *** ****** ***** 3 ******** ****** **** *** ***** taken ** *** ****** ***** ***** to *** ***** * ***** ********** (in *** *****) ** ***** ***** people ** *** **********: **%, **%, and **%.
**********, ****** ****** ** **** ** well:
*** ****-*** ****** ******** *** ******** based ** *** *****, ************* ******** multiple ****** ** ******** ***********, *** then ***** **** ** ***** ** similarity. ***back-end ****** **** ******** ****** *********** **** * ********** ** **** **** **%, and finally ******** ********* *** ******** ** *** ******* *****. [emphasis added]
Dahua/Tiandy/Uniview ********* ***** ** **********
*************** *****, ******, *** ******* ********* about ***** ********** ******* ******** **** real-time *********** ** **** ** ******* because ** * **** ** ****** to ******** **** *********, **:
the ******** **** ******** **** ** ****, *** *** ***** ** *************** ** **** ****. It is difficult *** * **** ** *** *** ******** **** ********; furthermore, the current face recognition is not 100% correct, and personal information will be leaked if it is wrongly posted, causing a series of problems, so many ****** *** ******** *** **** ******** *******. [emphasis added]
* ******* ******** **** ** "******** that *** ************** ** **** ********** is * *** '***********.'" *******, ***** Uniview *** ******* **** ****** **** these ******* ***********, ** **** *** do ****-**** ********** ** **** ** the *** ******. *******, **** ****** either **** ****-**** ******* ** *** "wall ** *****" (******* **************) ** manually ****** *******.
** *** ******, "****** ***** *********** and **** ******" *** **** ** identify *** *********, *** ********* ****. These ************ *** ********* ******* **** person *** ******** ** * ******** intersection ****** *** **** *** ******* was ***** ** *** ******, ****** police **** *********** ** ******* *** system *** ********* ********** *** *********.
No ************ ****** ** ******** *****
**** *** *** **** *** ************ claims ***** *** ******** ** ***** systems. *** **** **** ****** ****** is ****** ********* **** *** ******* themselves **** * ******** ****** ** mistakes. *******, ******* **** ** *% or **%, ***., ****** *********** ** not *********.
** *** ******** ****, ****** *** walking ******** ******* *** ****** *** for * ********** **** ****** ** time, ********** *** *********** ** ********* a ****-******* ***** *** ********.
** *** ******** ****, *** **** significant ******* **** ***** ** *** scale ** *** ******** ** ***** against. ***** ****** ********* **** ******** of ****** (*.*., ********, ***** *** the **** ** ***** *******, *** 20+ ******* ******). **** * ****** accurate ****** *********** ****** **** **** quality ******** ******, ****** ** ******** a ****** ********** *** ** ** million, ***** ********* ****** ********* *******.
* ****** ****** ***** ********* ****, as ********* ***** ******* *** *** 3, *, ** ** ******* (**** an ******** ****** ******) *** ****** ascertain *******, ******, *** *********** **** other ******, *** ** ** ** 20 *******, ***** ******* ** * legitimate ********* *** ****** *********** ******* in **** ***** ***** ************.
Providers / *************
********* ******* *** * ******** **** of ******* ********** *** ********* ********* "Smart *******" ********. **** ** *** providers/manufacturers *******:
- *************- **** ** ********, ******* ** IPVM's **** *****
- *****- ******* ***** **** ** *******, Anhui, ******, ********, *******, *** ***** provinces/municipalities
- *********- ******** ******** "********* **%"
- *******- ***** * ********** ** *******
- ******- **** ** *******'* ***** ****** in ****, * ****** *** ******** purchased ** ********* **** ** ******* ******** ******** ***** ******** *** ~$*** ***.
**** ** ***** ******* ********* ******* a ****** *********** ****** ***** * pedestrian ***** **** * ******** *** screen, ****** **** ** ***-**-*** ******, although ******** ** ** *********. ** long ** *** ******* ****** ****** recognition *******, *** ****** ***** ** relatively ****** ** *****.
Scale ****** ****** ** *****
***** ***** ******* **** ******** *********** attention, **** ********* **** * *** hundred ***** ************* ** *** ** China *** ***** **** ***, ***** they *** *********, *** ****** **** is ******, **** ** ********-***** ******* to ******* ****.
**** ** ***** ******* *** **** now ******** ** ** ******* **** a ******* ** ********* ** ******, including * *** **** * ******:
- *** ****-*********, *****, ********* ** ** *************
- *** ****-************, *******, * ***** ****** ** * intersection
- *** ****-******, *******, * *************
- *** ****-*******, ********, * ***** ******, * **** intersections ******** ** ******** ****
- *** ****-****** ********, *********, * *************
***** ****** **** **** ********* ***** systems, ****'* ******** ****** ** ******* of **** * ** ** ************* in * **** ** ********* ******* by ***** *******. ******** ** ** outlier ****** ************* ******* ** ** *******. *****, **** ** * ***** percentage ** ***** ************* ** * large ****.
*** *******, *** ****** **** **** the * ******* ************* ** ******* (population ~* *******). *** ****** *** shows *******'* ***** ******:
Punishment ******* ** ********
********** *** ********** ******* **** **** to ****, ******** *** **** ****** punishment ** ** ******* *** ******'* image ** * *** ****** ** that ********* ** *** "***** ** shame". *** ***** ** *** ********** and, **** *** *********** ** *****, is *** ******** ** ******** ** legal **********.
*******, * *** ****** **** ********** monetary **********. *** *******, *** ********* is ******** ** ** ********** *** 20 *** (~$* ***) **** *********** ******* ** ********* ****** ********. *** ********* *** **** go ** * ***** ****** ******* to *** *** ****, ** *** it ******* ******, ******, ** *** Shanghai ******* ****** ***. ** ********, in **** ****, ***** *** * report ******** ***-********* ************. *** *******, ** ******* *** more **** * ******* **********/**** (********* jaywalking), *** ****** ****** ***** *** be ********, *** ** ****** **** participating ** ******** *** * ******** license ***** *** **** ****. * similar ****** ** ***** *** ****** score ****** *** **** *********** *********. ***** ***** ** * **** received ** * ******* *** * few ******* ***** ** *********. ** reads:
[**** ****** ******** ******] (**** *******) on ****-**-** ** **:**:**, ** *** intersection ** ****** ********* **** *** Xinglong ****** *** **** ****** ********** (traffic *********). ****** ****** ****** *** business **** ** *** ****** ******* outside ** **** ************ *** **********. Contact ******: ********.
Appeals ********
** ********, ***** ** ** ******* process *** *********** *** *** *********** with *** ******'* ******** ** *******. In ************* **** **** ***** *** ******'* crackdown ** ********** (*.*. ** **** warning/second *******), *** ***** ****** ********** said *********** *** **** ** ****** if **** *** ***********:
** *** ***** ******** ** *** traffic ********* ** ******** ** *** electronic ******, ** ***** ** ** objection ** *** *********, *** ******** video *** ******** *** ** ******.
** *** ******** **** *** ****** penalty ********, *** *** ***** ** the ******** ****** ******** ****** ** the ******** ******’* ***** *** ************** reconsideration ****** ** **** ** ********* the ********; ** **** ** ************** lawsuit ** *** ***** ****** * months.
“***** ********" **** ** * *** Function ** ***** *******
***** ****** *********** *** ************* ** these ******* **** **** **** ***** building ** * *** ******* ** these ******* *** **** **** ******* is ********* **** **** ********* **** catching ***** ****** ****** *** *** jaywalked. * ***** *****************:
*** ******* ** *** ********** ***-***** capture ****** **not ** ******** ******** *** **** * *** *****, *** ** ****** ********* ** ***** ** ******* ****** **** and regulations through exposure, and then create a safe and civilized traffic travel atmosphere. Management is only a short-term measure. You have to curb related bad habits to be effective in the long-term. [emphasis added]
* ******* ****** ********** ** ******, Zhejiang **** ***** *********** ** ***** good ******, ******* ** ********* *****:
******* ******* ****** ***** ******* ************ to ******* *** ******* ******, ***** by **** *** ***********, *** ****** safely *** *******!
** ****** ********, *********, *** ***** pilot ****** *** ********* ** ********* 2020 ************ ******* ** ** ********** school. *** ****** *********************** ****** ***** *** **** **** ensure *** ****** ** ******** *** also ** **** **** ***** **** pedestrian ******:
*** ****** *** *** ***** ****-**-****** traffic ***** ****** *** ****** ** be ****** ** *** ******** ** the ****** ** ** **** **** account *** ****** ** ******** *** students ** *** *** ****, *** on *** ***** ****, ** ******* the ******** **** ** ***** *** to ********* ********'* ******* ****** *********.
Government ****** ***** ******* **********
**** ***** ** *** ****** *** local ****** *********** ** **** ******* of ********** ** ***** ******** ******** on *****, * ******* ***** ****** media **** ******* ** *******. *** example, ******* ******'* ******* ****** ** Ningbo, *************** ***** ** ** ****** *** redacted ******** *********** ** *** ******* on ******* *, ****. **** **** was ******* *** **** *******. ***** is ** ******* ** ** ***** from *** ****, ******* * ******* of *** *********, ** **** ** her ******** ****, ** ******, ****, and ******** ** **********.
****, *****'* ******* ******************* ******* ** *** ***** ******* on ******** **, ****, ******** ** the **** *******. **** ******'* ***** include ******** ******** ** *** ********* in *** *** ** **** ** the *********'* ******** **** *** ** number.
**** ********** ******** **** *** ******** Traffic ****** **** ** **** * website **** ***** **** ******** ** jaywalkers ********* ***** ***** *** ** numbers. *******, ** *** **** ***** down ***** ******** ****. ***, ** ** **** ****** for ***** ****** ** ******** ****** jaywalkers ******* ****** ****** ***** ********.
Netizens ********* **** ************* ** *******
***** *** ******** ** ****** ***** comments ** ***** ******* ******** ** IPVM **** ********, **** ******** **** concerned ***** ******* *** ******** **** "exposing ******" ** ******** ********* *** how *** ****** *** ****** **** areas.
******** ******* ** *** ******, *** they ****** ** ***** ** **** sure **** ** *** ******* *****, some ******** ****. *** *******, ** a ****** ***** * ****** *********, * ******* ******:
**** *** ** ***** ** ****** exposure? **** **** **** *** *** every ****, *** ** **** ***’* pay, *** **** ** * ****** black ****, **** *** ** ****’** still *******.
******* ******* ****** ********* ******* ** a ****** ***** ******* *********, ******* **** ****** ******** ** not ****** ** ***** ****** *** already **** ** *****:
***** *** ** **** ****** ******* and **’* ******* *******. ** *** think ***** ***** ** ****** **** about ****?
***** ******** **** ********* ***** *** the ****** ***** **** **** **** areas ** ********* **** ** ***** be ********** *** ******* ** *******, arguing, *** ******:
*** **** ** ***** ** ** short, *’* ** ********* *** * have ** *** ** **** ** across ** ****. **** ******* ** elderly *****? * ******* ********* *** law **** ******** *****, *** ****** solve **** ******* [** *** *** elderly *** ***** *** ****] *****.
**** ******* ** ***’** ******* ** the ************ *** *** ***** ***** red?
*** **** **** **** ** ******** is ******* *****?
Future *********
**** **** ******** ** ******* *** facial ********** ******* ******.