The Hikvision Smart Classroom Behavior Management System

By John Honovich, Published May 22, 2018, 08:53am EDT

Hikvision's rapidly growing offering of analytics, which we most recently examined with Hikvision's ethnic minority analytics, is now going into the education market where Hikvision aims to improve scholastic performance with artificial intelligence. This was first covered in the West by Reuters but we found and reviewed the Chinese source report.

In this note, we examine Hikvision's offering, the potential benefits and drawbacks, including accuracy concerns.

Classroom ******** ********** ****** ********

***** ***** ***** ********:

*** ************* ** ******** in *** ***** *** calculated ******* *** **** analysis ** ******* ******** improvement.

**** ******** * *********** of *** ******. *** is * ****** ** total ****** *********** ** all ******** ********** *** day, ** *** ******* below ***** ********* ****** neutral *********** *** *** angry ****:

*** ****** **** ****** what ********** ** ***** students *** *******, ******* their ***** ** ********* to *** ******. ** the ******* *****, ** student *** ******** ******* their ****:

*** ***** ******* ****** was ** ******** ** the ****** *****' ***** level, ** *********** / ranked ** *******, **** the **** ** *** right, ** ******, ***** the **** ******* ********:

********, *** ****** ******** a **** ** ******** who *** *** *******:

Camera / ******

* ******* ** *** camera / ****** ******* at *** ***** ** the *****:

Vote / ****

Accuracy ********

**************'* **********'* ********,********* ********* *************** ******* ** ********* glasses ** ***** *** marketing *****, *********** ******** ** a *******, **** **** so **** ********* *** emotional ***** ** *********. If ********* ****** ********** detect **** *********** ****** reliably (**** ** **** a *****, ** **** person ******* *******), *** risk ** *** ****** with ********** **** *** more ********* ** ****** by ******* ** ***** (e.g., ** * ****** sad, *****, ********, ** pain, ***.).

Behavior ****** ********

******** ******** ** *** system ***** ********** ****** student's ******** ** **** are ******** *** ****** by *** *** ****** interprets ***** ****** **********, which ***** *** **** of *************** *** ******* effects (*.*., ******** ************ their *********** ** ******* the ******).

Hikvision ******** *******

********* ********* ************ *** request *** ******** *******.

Future *****

**** ****** *** * pilot ** * ****** in *********'* **** ****. Where **** **** **** here ** *******. *********, it ***** ********* ***** to ************* *** **** innovation ***** ********* ** various *******, **** ** applications ***** ******** ******** are ****.

Update: ***** *****

***** ** *** ***** shared ****** ***** ** the ******. ** ** in ******** **** *** the ******** ***** ********* are **** **:

Comments (27)

Since we are on Chinese education, the topic of academic performance is likely to come up and the oft-repeated claim that China students are better than US ones.

For a contrary view from a Professor at China's top business school see Bloomberg's China's Top Economic Risk? Education article, key quotes:

headline-grabbing studies showing China's prowess on standardized tests evaluated only kids in rich and unrepresentative areas. When its broader population was included, China's ranking dropped across all subject areas.

According to the 2010 census, less than 9 percent of Chinese had attended school beyond the secondary level. More than 65 percent had gone no further than junior high. From 2008 to 2016, China's total number of graduate students actually decreased by 1 percent. Outside the richest areas, much of China's population lacks even the basic skills required in a high-income economy.

China is famous for requiring rote memorization of its students. But schools are also increasing classwork on communist ideology, Confucian thought and even Traditional Chinese Medicine. 


I wonder how well this tech would function when applied to a variety of people with different facial features than what a typical Chinese person possesses. I remember a facial recognition company a while ago telling us in a training class that they had difficulty training their system to differentiate different asian facial features as opposed to non-asian facial features. I wonder if these guys would face similar issue.

Either way im glad i didnt grow up in an age of video surveillance. How stressful I imagine that would be to be constantly evaluated.

I wonder how well this tech would function when applied to a variety of people with different facial features than what a typical Chinese person possesses


I am highly skeptical that this works even remotely close to what Hikvision is trying to portray here, so it would probably work equally poorly on students of any ethnicity.

Seriously, Hikvision can't even master basic cyber security concepts that have been well-documented for over a decade, and we are supposed to believe they are providing cutting-edge analytics like this?

I would not be the least bit surprised to learn that the "data" from this system is actually compiled by a subset of Hikvision's massive team of "engineers" manually viewing video and tabulating stats.


I look forward to telling a friend who is a high school vice principal, that now he can evaluate teachers using this tool and recommend firing any teacher whose students don't raise their hands enough, smile enough, or appear to be alert enough. 

The image shows all the asian kids being serious and the caucasion kid is goofing off. i am deeply offended by that.... J/K, i think its hilarious.

Back on topic, i dont get excited about all this bleeding edge technology anymore. Normally it simply doesn't work anywhere close to what customers expect. Nonetheless, someone or some company has to start somewhere with it.

This will be more added to their "Social Credit" system. 

I'm with you on this one. When I see it work, I'll believe it works. 

Very much against this. Don't get me wrong, I'm all for getting some discipline back in schools, but having this going in entirely the wrong direction. Respect and human interaction trumps machine learning by far.

Nonetheless, someone or some company has to start somewhere with it.

 When you have it, use it.


...which poses the risk of ... adverse effects (i.e., students manipulating their expressions to appease the system).

“Manipulating your expression to appease” is an essential social skill, practiced everyday starting with your first “Good Morning!”

Agreed people will always find a way to adapt against any sort of means of attempted manipulation of them.  So instead of kids using your energy to actually paying attention and focus  in class they will be using their energy to concentrate on how to appear to be focusing on class.  


A picture of the camera / system  from a different Chinese site:

Thankfully GDPR will prevent this to happen in European schools :)


Good luck Hikvision lol. Focus on cyber security and camera MTBF

ASIS should partner with Hikvision and deploy this inside their GSX learning sessions.


The system also displays a list of students who are not focused:

And we verified the other UI segment ranks students by how focused the system believes they are.

Chinese video clip where this is from here.

Does it matter if it works? To me what matters is that this use case was thought enough about to attempt it. Now think about all of the other possibilities that can be applied to this type of technology. Innovation comes from these basic beginnings. And sometimes innovation is scary. 

Data is powerful and lucrative. If there is a desire (demand) to identify trends, detect anomalies, etc., someone is going to try and build the solution (supply). Once the behaviors can be measured, they can be predicted. Maybe it won't be Hikvision, but eventually somebody may succeed at this.

Phillip K. Dick's precog psychics may have been a stretch but the idea of predicting behavior is upon us. Call Tom Cruise, reactivate the PreCrime unit.

Does it matter if it works? 

It matters if it works because there is going to be an increasing public pressure and ill-will if the public views these technologies as harming them and being fundamentally unfair (due to their inaccuracy). For the past few decades, the video surveillance industry has been viewed mostly positively since it overwhelmingly was used to target clear criminal acts and not unfairly targeting regular people. AI, especially applied cavalierly and inaccurately, could change that.

John, I agree. I am saying this matters even if it does not work. 

How well it works matters only if hikvision is concerned about their credibility and business ethics. It takes little effort to dream up some neat uses for technology, and only slightly more effort to mock up some demos. 

Any company can put out forward-looking claims like this, and many have in the past. If you present it as a technology demo, that is one thing. If you present it as being close to commercially deployable when it is not, then that is a sign of a company that is not ethically driven or concerned with their corporate image. 


AI is one of those things that are too useful to discard and too dangerous to take lightly ...

Under the guise of improving scholastic performance we are supposed to be watching a child emotions... to what purpose? Who's watching the watchers ...

As dystopian as it gets, creepy doesn't even begin to qualify it.

Thanks! That article cites SupChina's China To Curb Facial Recognition Technology In Schools which cites The Paper China: Department of Science and Technology of the Ministry of Education: Campus promotion of face recognition technology should be cautious, will limit and manage. One notable quote:

The collection of the use of minor information should be approved and authorized by the guardian.

We'll look to see if we can get more information on this.

Define “guardian”.

Is there any reason to believe that it isn’t the same definition as it is in the west? A guardian is the parent or other responsible adult of a child.

forget face rec, China goes direct to the source

A headband that claims to monitor children’s brainwaves in order to improve their focus is available for purchase in China, aimed at “Tiger” mothers and fathers who’ll do everything to help their children succeed. But many of its target customers are already creeped out.

The Focus1, or Fu Si, headband, from US-based startup BrainCo, claims it can measure how closely students are paying attention through electrodes that detect electrical activity in kids’ brains and send the data to teachers’ computers or to a mobile app. A light on the headband that gleams red, yellow or blue, also purports to signal how engaged a child is with the task at hand—with red being the highest level of attention.

This could be brilliant or a complete failure. Need i remind anyone of the great “suck cut” debacle in 1992?

Read this IPVM report for free.

This article is part of IPVM's 6,735 reports, 909 tests and is only available to members. To get a one-time preview of our work, enter your work email to access the full article.

Already a member? Login here | Join now
Loading Related Reports