Video Analytics Manufacturer Performance Guide

*******

********* ******** ****** *** most *************, **** ************* deep ******** ********* ********** strongest, *** ***/********* ********* suffering *********** ******. ***** most ****** ************* **** some ****** **** **** well, ***** *** *** exceptions, *** *** **** in ****.

***** ** ******** ** hours ******'* ********* ******* ** the ***** ******** *********, **** ****** **** average ** **** ********** analytics, **** *** *** performers ***** **** ******** algorithms.

Included ********* ** ***-**

***** ************ ********* *** typically ******** **** *******/*********, or ** * ********-**** add-on. ***** ************* ******* analytics ****** *** ****** or ******** *** **** to ************* ******** ************. However, ********-**** ********** ****** subsidize ** ****** ***** analytics **** *******, *** commonly **** ********* *** tens ** ******** ** dollars *** *******. ** such, ********-**** ********* *** at * *********** ************ compared ** *** ******* video ************ *************, ***** all ********* **** ******* and *********.

*** *******, ******** **** offer ** ******* *** include ****** *********** *********** for ****, ***** ********-**** providers **** ******** ** dollars *** ******.

************, ***** ** ********** complexity **** ***** * 3rd ***** ******** ***-** in * ***, ******** to ***-**-*** ******/*** *********.

Camera *************

********

******** *** ************** *** end-to-end ***** ************ ******** by ******** ** ******* integrated ******* *** ***. However, ********'* ********* ******* are ********** ********* ******** to **** ********* *** they *** * ********** dealer-only *******.

********* ******* ****** ****, ******** ****** ******** series ** **** ******** analytic *******, ************, ***** ***. *** *********** ** Avigilon ********* ****** **** between ****** ******/******** ******** than **** ************ *********. IPVM******* ** ******** *** analytics***** ****** ******** ********* classification ** ******, ********, bikes:

**** *** **** ****** at ******** ****** ***** alert ******* **** ****, headlights, *** ************* **********:

*** **** ******** ******* accurately ******** ****** ******** including ******** *** ******* colors, **** *********, ***.

***** **** ****** ******** in *** *** ***, Avigilon ****** ******** ********* including********** ******,****** ***********,******* ******** *********, ********* ********** ************** *** *** ***** in **** ***********' *********.

*****

***** *** ********** ***** to *** ***** ********** analytics, **** ***** *** party *** *******. ***** Bosch's ***-********** *** ********* are ******* ** *** most ********* ****, ****, and **** ******, **** also ***** **** *********** *********** *****-**** ***** *** 4000i *******, *** ***-****** 5000i *** *****.

**** ******* *** ***** Bosch's *** ****** **** accuracy *********, ************** ** people *** ******** ******* types, ******** *****, ****, and ***** ********, ** well ** ****** ***** alert **********.

*******, *** *****-********** *** suffers **** ****************** ** people *** ********, *** false ****** ** ******* and ****:

************, *********** ** ***** analytics *** ************ **** complex **** ********, ********* manual *************, **** ****** often ********* ** ************.

*****

***** ** *** *****'* second-largest ***** ************ ************ and ******* ** *********, includes ********* **** **** of *** ** *******, however, ***** *** *********** issues **** *****'* ******** intrusion ********* ** **** of *** ******-**** *******. Dahua's ******* *** ************* lower **** **** ***********. For *******, *** **** learning *********+ ******* **** 50% - **% **** and ***** ******* ** better *********** **** ********* DeepinView.

*****'* *********+ ****** ** a *** ********* ** our *****, ******** ***** alerts **** *** **** common ******* ******* *** outdoors, ********* *******, ****, shadows, *** ********.

*****'* ***** ******* ********* perform **** *** ***** alert ** *******, ************* humans, *** ********, *** its ******** ** ********* only ** *****'* *** platforms (****/*** *******).

*******, **** ** ***** analytics *** **** ******** than ***** ***** ****** Detection *** ******** ********* analytics, ***** ****** **** numerous ***** ***** ******:

******

****** (******** *******) ******* have *** ************ **** positioned **** ** ******* to *********, *** **** emerged **** *********** ** ***-**** competitors**** ***** *** *********. Hanwha ******** ************ * ** ****** in ****,****** **** ******** ****** and ******* *********. *******, deep ******** ** **** supported ** * ******* in *** ***+ ** camera ****.

******'* ******* * ** cameras *** ******* ** other *** ********** ** our *****, *** **** minor ***************** *** ***** alert ******:

************, *** ******** *** accurate *** ********** **** multiple ***** *** ******.

******'* ******* * ****** detection ** ******* ******** to ***** ***-**** ******** analytics, **** ***** ****** on ******* *** ****** outdoors, *** ***** ********* analytics ******** **** ********** false ****** ** ****, foliage, *** *******.

*********

********* ** *** ******* video ************ ************ *** ships ********* ** *********** every ** ****** ** offers. *******, *** ********* detection ********* ******** ** their ******** ***** ** poor. ************, ********* ****** multiple ***** ** **** learning ******** *******, ****** recognition, *** ******* ***** cameras. ***** *** ********** less ********* **** ******* brands.

*********'* ******** ** **** learning ******* ***** ******* accurate ********* *** **************, compared ** ***********' **** learning *******

*******, ***** *** *** 2.0 *** ********** **** suffer **** ****** ** animals, ******, *** *******, though **** ********** ********** humans *** ********.

*********'* ******** *** ********** analytics *** * ******* improvement **** *** ****** cost *******' ********* ********* analytics, ***** ******* ** most ***** ***** *******, only * ***** **** up **** ******** ***.

****

*** ******* ******* ** camera ************, ************ *** ***** **** learning ********* ** ****, *** **** ** the *** ********* ** do **. **** ****** Analytics ********* **** ******** a ******** ** ****' overall ******* *** *** top-performing **** ******** ******* on **** * ****** out ** ***.

**** ****** ********* ******* on ****-******** ********** **** equipped *******, **** ** the **** ***** ** III ****** *** ****** and ******* ********* *********** in *** *******, **** strong ***** ***** ********* and ******** *********.

*******, ****' ***** ********* – **** ******* ******** Object *********, ********* ********, Motion *****, *** **** – ****** **** ******** false ***** *** ***************** issues:

************, **** ** ****' analytics ******** ****** ******** found ** ********, ****** AI, ** ***** ***.

*******

*******'* **** ******** ***** Motion ********* ** ********* on **** ***** **** of *** ** *******, and ***** ****** *********** in *******, **** ** false ****** ** ****** sources ******* ** ***.

*******, *** ******** ******** additional ************* *** **** accuracy *** ******** ****** only, *** ********. ************, Smart ****** ********* **** not ****** ******** ********.

***** ****** ********* ****** are ********** **** ********* XProtect, ******* ******** ******, and *******'* *** ***** VMS, *** *** *** supported ** ***** ***-***** VMSes.

*******

******* ** *** #* camera ************ ** *****, a ***-**** ******** ** the **** **** ********* poor ******** ***********. ***** Uniview ****** ********* ********* on **** ** *** camera ******, *********** ** poor/below *******. **** ******** deep ******** ********* ** 2020, ***** ********* ********** worse **** ********* ****** Dahua *** ********* *******.

******* ***** **** ******** analytics ******** ***** ****** on ***** *******, *** wheels, *** ****** ****:

******* ******** ********* ********* also ******* ******, **** false ****** ** **** common ******* (*.*. ********* from ***** ** *** Mode):

Hikvision *** ***** - ** *** ** **********

*** *** *** ************ of ********* *** ***** hardware******* * *********** ********* for **** ************* ******* of *****. ********* ********* growth ***** ******* **-**** ***** **** ** the ******, ********* *** ***** were ******** ********** ** offer ******-******, ******* ******** cameras.

*******, ** *** **** since *** ** *** and ********* *** *******,********* ******** ********** ****** in *** ** *** Europe*** ********** ********** ******* ********* while ****** ***** ****** abuses.

VSaaS ****** *************

***

*** ****** *** **** performance ** ***** ****** analytics ** *** *******, avoiding ***** ****** **** all ****** *******, **** as *******, *******, *******, rain, *** ******.

*** **** ********** ********* issue ****** *** ***** was *** ***************** ** inanimate ******* ** ****** or ********, **** ** a *******, ******, ** dumpster.

*** ****** ***** ********** clothing *** ******* ***** search *** **** *** offer ****** *********** ** LPR, ***** ** ****** with ***** *******.

************, *** ****** ******-***** deep ******** ********* *** 3rd ***** *******, ***** is ******** ** **** space.

*******

******* ********* *** ****** as ******** ** ***-********** traditional ************ ****** *************, with ****** ****** *** vehicle ********* ** **** ranges, ******** **** ***** alerts ****** *** ***** rain.

******* **** ****** ******* Behavior *********, * ***** analytic *** ******* ** most ***********, ***** ***** well ** ******* ** identify ******-***** ******** *********:

******* **** ****** ****** recognition ********* ** *** cloud **** *** ********** subscriptions, ***** *** *******, and ******** ***************** **** changing ******** *** ****** color *****.

*******

*** ** *********** ****** controversies (******* **** ****,******* ******* **********), ******* ** *********** the **** ********* **** for ********** *****. **** have ******* ** ********** deep ******** ********* ** all ** ***** ******* and ** *** *****.

******* *********' ******* ***** is *****-*****/**** *** ************, ~25 *** *** ****** detection, ***** *** ******* PPF ******** ** **** competitors.

************, ******* ******* **** false ****** ** ****, which ** ****** *** many *********.

******* **** ******** ****** recognition, ******** *****, ***** analytics, *** ****** ********* with ***** ****** ************. However, *********** ****** ************* based ** ********, *** angle ** *******.

Software-Only ******** ******** ********

***** *** **** ****** of ***** ******** ********-**** providers ****** *** ***** (********* ** *** ***** Analytics *********). ***** **** ***** common ***** ************ ********* (e.g. ****** *** ******* detection), **** ***** ** specialized/niche ********** (*.*. ****** recognition, *** *********, ***).

***** ********* (****** **************) are ******** ** ******** used ********-**** *********. **** is *** ***** ******** on *********** ** ********, as **** *** *** tested **** ********-**** *********.

*********

* **** ****-****** **************** ************* **** *********** *** surveillance, ****** *******, *** mobile ************. **** ****** $43 ******* ** **** and***** * *** ***. ******** ** ****** at $*,*** *** *** channel, *** ********* *** cost ** ****** ********. Additionally, ****** **** *** required, ********** **** ******** to ******* ************ *******.

*********'* ****** *********** ** able ** ********* ******** at ~**-** *** ** well *** ******, * lower ***** ******* **** many ***** *********, ***** requiring *** *** ** more.

*********'* ***** ************** ********* was ****** ** ***** watchlist *******, **** false ****** ****** *** cameras, ***** ***** ******* confidence ********* ** *.**, with * **** ******** as ***** ** ***, with **** **** *** subjects ******* ** ***** of ***** *******. *******, this ** * ***** population ******** ** **** public ****.

************, *********** ******** (*.*. high ***, *** *****/**** and ** ***********) ****** moderately ******** ******, **** missed **********, *** ********* algorithm **********:

*********'* ********** *** ********* has ********* ******, **** with *********'* *** *******/***** clients ** **** ** integration ****** **** *****-***** VMSes, ********** **** ******** to ****** ************ ********* (e.g. ********)

********

*** ** *** **** commonly **** ********-**** ********* in ***** ************ *********, outside ** *****. ******** offers * ***** ***** of ******, ******* *********, and ****** ***********. **** were ******** ******* (***** ** ********* and ****) ** ****.******** ** ********* **** in ***** *******, **** a *** ******* *******, for ~$***-*** *** *** channel (~$**,*** **** ****).

******** ****** ******** ***** in *** ***** ******** platforms, **** ***** ******** and ******** ********* ******** search ************* ******** ** typical *** ******, **** features **** ** *****, clothing *****, ******* ****, direction, ***., ******* ************ well ** ******* *** moderate *** ***** ******.

*******, ** ******** ********** *** ***** ***** recognition ******** ******, *** ** ********** causing ******** ** ****** red:

************, *** ****** *********** analytics **** *********** **** others ** **** ******, such ** ********* *** Avigilon, **** ** ***** recognitions ** * ***** of *******, *** **% lower ****** *** **** required *** ******** ***********:

*******, ****** *** *** recognition *********** *** ****, even ** **** ***, narrow ****:

******** **********+ ******* ******** *** ******* video *** ******* ****** (including ******* *** *********) support **** ********. *******, in **** ** ********'* VMS ************, ******** *** performed ** ** ******** Briefcam ******, *** ** the ****** *** ******.

*******

****** / ** *********************** **** ** ********** ***** * ***** range ** ******, *******, bag, *** ********* *********, with *** *** ******* integrated ***.

******* *** ** ********* street ***** ** ~$***-*** USD *** ******* *** live ********* (********* *********, facial ***********, ******** ******, etc.) *** ~$***-*** *** video ******, *** *******. These ****** *** ************* lower ************************.

******* ****** **** ****** person ********* *********, ***** facial *********** *********** ** well-lit ******, **** ** harsh ******. ******* ****** detection **** ******** ******** and ****** ********* **********:

*******, ** ******** **** common ***** *********/****************** ** raindrops ** ******* ***** and ** ***** ******* which *** ** ** issue ** **** ************.

*******'* ****** (*********, **** rec, ***.) *** ****** are *** ********** **** third-party *****. ****** *** be **** ** ***** using ****, *** **** configuration **** ** ********* manually. ******* ******** ** many *** *** **** to ******** ***** (**** and ********) *** **** limited *-*** ******** ***********.

Challenges ********** ******** ****

********** ********-**** *********' ********* capabilities *** ******** ** complex, ******* ** ********/*** requirements, ******* *** ***** VMS ************, *** ***** vertical-specific (*.*. ****** ******, healthcare, ********) **-**-****** ********.

SoC ************ *** *********

****** ************* *** *** party *** ********* ******** analytics ** ************* *** justify ******* *******. ****** sophistication, ****-********-***** ********* ** driven ** ************ ** SoC **** ******** ******* and ************ ** ******-********** neural ********.

*** ********** **** *** 2021 ** *** ********* for ******* *** *********. The *****'* ******* ***** surveillance **** ********,****** ********** *********, *** ******* ********** **** **** in ********* ****. ** *** ******** side *** ******, *********** inventory *** ***** **** reports ********** **** ***** manufacturers **** * **** or **** ******.

Significant ******* ********** - ***** *** *** *****

***** ***** *** ********** significant ********** ** ***** surveillance ** ****, ** is *** ***** **** companies/trends *** ***** ** break ***.

**** *** **** ** millions ** ******* ** to ****** *********** *********, however, ** *** ****** controversial **** ******** ***** over *** *** *** the ******** **** ** create **.******** *.*. ****** *** cities**** ***** ********** ** limit *** *** ******* ** *** *.*.****** ************ ** **** recognition.

*******

** ****** ********* ****** to ** ****** ** camera *************, ** **** release ****** ******* ****-********-***** analytics ** ****, **** a ******* ******* ** AI ****** *********, ************ in *****-**** ******.

** ** *** ***** how **** ****** ******** analytics **** ****** ************, but ******* ** *** shift **** **** ********** purely ** ****** **********, we ***** ****** * similar ***** **** ******* and ** *********.

*** ********-**** ******** *********, it ** *** ***** if *** **** ****** mainstream/mass-adopted. **** *** **** successful ********-**** ********, ********, is * *****, ********* system, **** ********* ** large ********** *****.