Startup: Face Recognition Intercom Chui / Now Trueface AI Software Analytics

By Brian Rhodes, Published Mar 13, 2014, 12:00am EDT (Info+)

This is one of the most ambitious security / surveillance startups we have seen in years.

Face surveillance startups, targeted at the professional or government markets, have struggled, again and again.

Now, a new company, Chui, is combining facial recognition, a video intercom and a doorbell in a single device, acting as a virtual bouncer - lets those you approve in, but dismisses those not on the list with a custom audio message.

Is there a market for these advanced features, and at what price? We answer those questions in this note.

2019 ****** - ***** ** ********-**** ********* / $*.******** *******

**** ********* ********** **** ****, *** ******* **** **** smart ******** ********* ** ** ** software ***** ********* *******. ******** **** they *** *** ** *** **** or ** *** *****, ********* ****** GPUs *** ********** ******** ****** *************:

IPVM Image

**** ******** $*.* ******* ** ****** A ******* ** **** ****, ******** their ***** *********** ** ~$* ******* since ****.

**** ** * **** **** *** crowded ***** (********* ** ***** *********) ** ******* (********* ** *** ***** ********* *********), *** ***** $* ******* ** good *********** *** ***** ************, ** is ************ ***** **** ***** ** analytics ******** (** ******* ********* ****** $** ******* (Now $** *******) *** ** *****).

** **** *** ****** ********, ** we ****** ******* ** *** **** their ********* *******.

Product ********

**** ***** **** ********* ******. *** promo ***** ***** **** * ***** minute ***** ** *** ******:

Key ********

**** **** *************:******* ** *** ****** ** *** claim ** ************ **** *** *****'* mobile ****** **** ** *** ** latency. *** **** ** **** ******* rings *** ****, *** ****** ** able ** ***** ***** *** ***** remotely ******* *** ******** ****** ** carry ** ***-*** ***** ************ ******* parties, *** **** ***** **** ****'* integrated ** ******

******** ***** ********:** **** ** **** ********** (** doesn't ********) *** ****** ******** ** front ** **, ** **** **** a ****** ***** ******* ** **** individual. *** *** ***** **** * welcoming ******** ** ******* ** ****** to "*** ****" *** ******* ** unrecognized **********. *** **** ***** * running *** ** *** ***** ** gained ****** *** ** ********* ** the ***/****** ****:

IPVM Image

********* ******:**** ****** ** **** ********* *** of *** *** *************, ** **** ** * '********' face ** ********** ** **** ************* unlock *** ****.

****: *** ***** ********** *** '***** doorbells' ******* ** ** $***. **** is ****'* ***-********** *****, *** **** the **** ** ******* ***** *********************. *******, ** **** ******** ** it's ******** ********, ** **** ***** more ***** *** *** **** *****.

Doorbot **********

**** ******* ********* *** ******** ** Doorbot (*** ******** ****), * **** **** *** ***** of ****, *** *** ****** **** in *** *****. **** ***** ***** to ***** ****-**** ****** *********** ******* the **** *** ****** ****** *** integration ******* *** ***** **** *** door *****.

*******, **** ****** ** *** **** than **** **** *********** ** ***** unit. *** **** ***** *** ******* to *** **** * **** ****** feed, ** "** ******", *** * commercial ******* *** ********* '**** & Attendance' *** *****-****** ************** ********* ** Code *******.

Sketchy **************

**** ** *** * ****** ******* yet. ***** ** *** *********** *****, the ******* ** **** ** ****** that **** ***** ************. ***** *** project ** ** ***** ** **** it's ******** ****, ***** *********** ******* is ** ********* ** ******* ***********. Additionally, **** *** **** ********* * raise ** $**,*** ***** ** *** much, **** ******** ** ***** ************* campaigns.

Potential ********

**** **** *******'* *******, **** *** most ********* ******* *** **** ****** to **** *********.

**** *** '*******'**********, ********* ** **********. *** **** does *** ********** **** **** ** be ** *****, *** * ******* of ***** ********* **** *** ********, subjects ******* ****** *********** **** ********** or ****, *** ***-***** *********** *** kill * ******* ****. ********** *********** Chui ** * *** *******, ** do *** ****** ****** *********** ** match *** *******'********* "**.* ********" *****.

**'* ********* ***** * **** ****** especially ** ** ** ****** * novel ******** *** *** *********** ******.

Comments (16)

They couldn't think of a better name than Chui??

Very interesting and eager to see if this technology will actually work like they say it well. I have heard some complaints with Doorbot and they also made a lot of promises before that product came out. I have not seen Doorbot even try to break into the security market, and I wonder if Chui will do the same thing. Companies such as Aiphone or Comlit should think of buying one of these companies out.

This interview explains the name: "They say the unusual name (CHUI) is derived from the Swahili word for “leopard” – an animal known for its agility and adaptability. They feel those traits have served them well as they move forward."

The pitfalls of Doorbot could be repeated by Chui, but at this point that level of detail is pure guessing.

2D Face for access control? A disaster waiting to happen......all you need is a photocopy of an enrolled face with enough pixels and bingo the door opens......there is a reason why Face Detection is not a common application for access control as single token authentication.

As far as I know, face recognition technology already found out how to handle a photocopy, not a real face of a real person. Only problem is identical twins. But I am also wondering the level of face recognition technology CHUI is adopting.

Which face recognition 'technology'? All face recognition vendors?

Btw, undisclosed B has real experience in this domain and is not simply throwing out conjecture.

I recently had a chance to talk with one of face recognition manufacturers and the first question I posed was like that. They showed me a demo to explain how they implemented the countermeasure to exclude a photocopy etc. Additionaly, in another meeting with IRIS recogniton company they also mentioned about the similar countermeasure so dead eyeballs could not 'trick' the system contrary to the movie, "Minority Report". Anyway, as you mentioned, it won't be all vendors. Let me check if that was 2D or 3D.

Jeffery thanks for taking the time to respond to my post, as John mentioned I do play in this space and while I am not a Research Scientist I do consider myself a SME. I welcome more players to the space to make it more mainstream but its unfortunate most potential users as looking for too many 24/CSI type applications.

While most folks look to Face technology to ALERT in real time on a watch list in a uncontrolled environment its just not practical in the real world today. We all want the technology to work like it does on TV but reality is there are just too many variables in the 2D Face space that I am playing in. The technology is there when it comes to computer vision but sometimes folks are sort of just enamored by the WOW factor and expect to much in certain applications.

A few things that are challenges in my past experiences between Facial Recognition and Facial Detection/Surveillance might help explain my thought process. Most users seem to want unobstructed Face technology and while it can work it takes money, time and resources.

  • Low Resolution Images vs Higer Resolution Database Images for Alerting
  • Uncontrolled/uncooperative Subjects
  • Uncontrolled Environment
  • Camera/Sensor positioning

3D Face technology on the other hand is another story but comes with a cost and can still have mixed results, you get out of it what you put into it. I think the industry as a whole has positioned Face incorrectly, even where I am has it challenges and many times I and trying to reset expectations when its pitched wrong by partners.

Now a somewhat controlled environment (static subject/camera positioning)like the Chui applications have a great place in the residential and small office market for access control, I wish you the best of luck in your ventures.

I am not suggesting the effectiveness of this one way or the other, but the developers say they have protected against photo spoofing in some manner (from the FAQ):

If they say so, I believe them! wait, what...

That said, the bigger practical issue will be it not recognizing people due to lightning issue as well as people having to take off their sunglasses, hats, etc. See if they have any 'innovative computer vision techniques' to deal with that.

Thanks for clarification and link Brian, I am going to see if I can get my hands on one and check it out. Regardless of the outcome of my posts I still think this is a affordable yet feature rich intercom and cool technology for stand-alone access control. Chui states that they can use the FIR as Secondary Layer and that’s good enough for me whether it’s a QRCode or some other token.

I am curious to see just how far 'innovative computer vision techniques' (which is like saying 'made of space-age material!') takes them. I suspect any claim long on fluff and short on detail is dicey at best.

I doubt it holds up to the "Dwight/CPR Dummy" spoof:

man you guys at too funny, I should have not waited years to start playing along......

So I tweeted one of the principals at Chui for more details on the 'picture spoof' protection. Here's the response:

He essentially says that the image is qualified as authentic by the presence of 'micro movements' and depth, which usually corresponds to light contrast across the subject.

If that helps, great. As usual, 'caveat emptor' applies.

So then would a recording of a facetime session played back in front of the unit be enough "micromovement"?

Agree
Disagree
Informative
Unhelpful
Funny

Update: original campaign said shipping December 2014. However, it's June 2015 and the current website indicates that it is still not shipping. I could not find a clear indicator when it would ship.

Agree
Disagree
Informative
Unhelpful
Funny

2019 Update - Pivot to Software-Only Analytics / $3.7Million Funding

Chui rebranded as Trueface AI in 2017, and pivoted away from smart doorbell solutions to an AI software video analytics company. Trueface says they can run at the edge or in the cloud, utilizing NVIDIA GPUs for supporting existing camera installations:

image

They received $3.7 million in Series A funding in July 2019, bringing their total fundraising to ~$5 million since 2017.

This is a move from one crowded space (Directory of Video Doorbells) to another (Directory Of 109 Video Analytics Suppliers), and while $5 million is good fundraising for video surveillance, it is considerably lower than other AI analytics startups (AI Startup Anyvision Raises $28 Million (Now $43 Million) Led By Bosch).

We have not tested Trueface, so we cannot comment on how well their analytics perform.

Agree
Disagree
Informative
Unhelpful
Funny
Login to read this IPVM report.
Why do I need to log in?
IPVM conducts reporting, tutorials and software funded by subscriber's payments enabling us to offer the most independent, accurate and in-depth information.
Loading Related Reports