TVT / InVid Facial Recognition Tested

By: Rob Kilpatrick, Published on Mar 25, 2020

Facial recognition is frequently sold for thousands of dollars per channel but some China manufacturers are offering full facial recognition systems for under a thousand dollars.

IPVM Image

But how well do such low-cost facial recognition systems work?

We bought TVT OEM InVid's facial recognition system to see where it worked and where it did not, including:

  • Does it miss people walking through the scene?
  • Does it misidentify people?
  • Does confidence drop in different scenarios?
  • Does it recognize inanimate objects as faces?
  • What PPF is required for proper recognition?

For our full report on the face rec camera used in this test: TVT / InVid White Light Camera Tested Vs Hikvision ColorVu

****** *********** ** ********** sold *** ********* ** dollars *** ******* *** some ***** ************* *** offering **** ****** *********** systems *** ***** * thousand *******.

IPVM Image

*** *** **** ** such ***-**** ****** *********** systems ****?

** ****** *** *** InVid's ****** *********** ****** to *** ***** ** worked *** ***** ** did ***, *********:

  • **** ** **** ****** walking ******* *** *****?
  • **** ** *********** ******?
  • **** ********** **** ** different *********?
  • **** ** ********* ********* objects ** *****?
  • **** *** ** ******** for ****** ***********?

*** *** **** ****** on *** **** *** camera **** ** **** test:*** / ***** ***** Light ****** ****** ** Hikvision *******

[***************]

Small ********** ******

**** **** **** **** was ********* **** * small ********** *** ********* size *** ** *********** restrictions. ********* **** *** <10 *** ***** ******** observed **** **** ***. Performance ** ****** *********** will ****** ********.

Executive *******

*****'* ****** *********** ******/**** offered ***** *********** ** well *** ****** ** our **** ** * very ***** ********** *** may ** * ********** option *** ***** **** tightly **********, ****** ****** of ****.

*******, *** **** *********** scenes ***** ******** ** less ********** (*** *****, WDR, ** **** ****** of ****), ** ***** hats, **********, *** ***** cannot ** **********, ***** is ****** ** ****** from ******** ******.

************, *****'* **** *** requires ***** ******* *** does *** **** **** third *******. ** *** positive ****, ***** ****** are ****** *********** *** face *** *******, ~$*** or ****. ** *** negative ****, *********** ******** **** ***********, **** ** *********.

Key ********

***** ****** *********** *** the ********* *********:

  • ***** *********** ** **** Light:***** ****** *********** ********* well ** *** **** light ******* (~*****), ********* faces ** ****** ******* directly ****** *** ****** even ** ****** ******** heights ** ~**' *** downtilts ** ~**°.
  • *** *** ******** *** Recognition:***** **** *********** ******** and ********** ***** ** ~94 ***, ***** **** several *********** **** *********** systems ***** ******** ~*** PPF *** ****** ***********.
  • ****** ****** ** ****** Individually ** ****:****** * ****** ** a **** ** ****** and *** ** **** from *** *** ** taking * ******* ** them ** ** ****** in **** *** ********.

*******, ***** **** ******* negatives:

  • **** **** ***********:*********** ** **** ****** (~0.02lx) *** **** *** to *** ***** ***** LED ************ ******** ***** in *** ***** ** view, ***** ****** ******** misses ** ********** ******* dropping ** ~**% **** faces **** ******** *** recognized.
  • **** *********** ******* **** Accessories:********** ******* ************* (~**%) when ****** ****** ******* the ***** ******* **********. Wearing **** ********** *** a *** ******** ** complete ******.
  • ** ********* **** ***** Half ** **** *******:***** ****** ** ****** people ******* ******* *** scene **** *** ****** half ** ***** **** covered, **** ** ******* wearing * ******* ****.
  • **** *** ***********:********** ******* ~**% **** subjects **** ******** *******, such ** ******* **** glass *****/*******.
  • ** ******** *********:******* * ******* ** a ******* ** ** the ****** ******** * positive, **** ********** *****.

No ********** ** *** ***** / ********** ***********

****** *******, *****'* ***, we ******* **** *** PN1A-16X16F *** ** ********* for ******* (** *********** by *****) *** ***'* first ********** *****, *** the ***** ****-**** *** their ***** "***** **********."

*******, ***** ******* **** models **** ** ****, we ***** ** ********* difference ** ***********. ************, InVid *** ******* ** any *********** ********** ******* the ***, *** **** us **** *** **** difference ******* *** *** was ********* ********.

Solid *********** ** **** *****

** *** *****, ***** performed **** ** **** light, ********* *** *********** faces **** ** ******* downtilts ** ~**°.

IPVM Image

Poor **** ***********

** *** **** ***** the ******* ***** ***** LED, ******** ****** ******** due ** ************ **** the ****** *** ****** to *** ****** ***.

IPVM Image

******* **, ************ ** reduced, **** ***** ********, but ********** *** ~**% lower **** **** *****, resulting ******** ******.

IPVM Image

****: ***** **** ******* suggested ** *********** ****** **** ***** but **** *** *** make * ********** ** performance ** *** *******.

Poor *********** ** *** *****

****** ******* ********** ******* ~30-40% ** * ******* scene **** ** *** entrance ** * *****, though *********** *** ***** possible ***** ******* ********.

IPVM Image

Low ********** *** ******** ****** ******* **** ***********

******** ******* ******* *** scene ******* ********** **** often ****** **** ** low ******** ******* (~*') and ******* ********* (~**°).

IPVM Image

********** ******* ** ~**% when ******* **********, ********* in ******** ***** *********.

IPVM Image

*******, ****** **** ********** missed **** ******* ******* the ***** ******* ********** and * ***.

IPVM Image

No *********** **** ***** **** *******

****** **** *** ********** at *** **** *** lower **** ** ***** face *******, *** ******* wearing ******** ****.

IPVM Image

Failed ******** *********

***** ***** ******** *********, producing * **** ********** false ******** ** *** picture ** * ******, near ***% **********.

IPVM Image

Lower *** ******** *** **********

** *** *****, ***** recognized ***** ** ~** PPF, ******* ***********'* **** **** **** recognition ********* ******** ********* ************ **** ******.

IPVM Image

Face ********* ******* *** ****** ***

****** ** ********** ** the **** *** ** done **** *** *** and **** ******** *** to **** * ******* of *** ****** ** upload *** **** **** phone, ***** ******** ***********, and ****** ******/*****.

*******

*** ***** ****-****** ***** for~$*** ********* *** ***-************** ****** sells****** *** ~$*** ***.

Versions ****

*** ********* ******** **** used ****** *******.

  • ***** *** - **************: 4.3.0.0
  • ***** **** - ******: 1.4.2.30787B191012.N4I.UI(******).**

Comments (0)

Login to read this IPVM report.
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.

Most Recent Industry Reports

OnSSI Founders Return, Start Corsight on May 25, 2020
The OnSSI founders are back, less than 2 years after selling OnSSI to Qognify, they have returned to Corsight, a spin-out of an Israeli AI...
GeoVision Presents AI and Facial Recognition on May 22, 2020
GeoVision presented its AI analytics and facial recognition at the April 2020 IPVM New Products show. Inside this report: A 30-minute video...
Density Presents Occupancy Monitoring For Coronavirus Protection on May 22, 2020
Density presented its cloud-based occupancy sensor to deal with Coronavirus at the May 2020 IPVM Startups show. Inside this report: A...
Openpath Presents Two Door PoE Controller on May 21, 2020
Openpath presented its new PoE controller at the May 2020 IPVM Startups show. Inside this report: A 30-minute video from Openpath including...
Bosch Presents MIC 7100 Extreme PTZs on May 21, 2020
Bosch presented its MIC 7100 Extreme PTZs at the April 2020 IPVM New Products show. Inside this report: A 30-minute video from Bosch...
Hikvision Chairman Targeted For Sanctions As Federal Watchdog Calls Out Hikvision "Serious Religious Freedom Violations" on May 21, 2020
The US government's religious freedom watchdog has criticized Hikvision for being "credibly implicated in serious religious freedom violations"....
Hikvision Temperature Screening Tested on May 20, 2020
Hikvision has ramped up the promotion of its 'temperature screening' system, including their salespeople arguing for no blackbody needed. But how...
Axxon Presents VMS 4.4 and AI Behavior Analytics on May 20, 2020
AxxonSoft presented its VMS 4.4 and AI behavior analytics at the April 2020 IPVM New Products show. Inside this report: A 30-minute video...
Indoor Robotics Presents Tando Aerial Drones on May 20, 2020
Indoor Robotics presented Tando indoor autonomous drones at the May 2020 IPVM Startups show. Inside this report: A 30-minute video from...
Directory of 89 Video Surveillance Startups on May 20, 2020
This directory provides a list of video surveillance startups to help you see and research what companies are new or not yet broadly...