Hikvision is marketing "safer, faster, smarter" with their Fever Screening Thermal Solutions, (renamed last week to Temperature Screening Series), their offering form the booming "coronavirus camera" market.
In this report, we look at Hikvision's fever screening series, their setup and accuracy claims, and how it compares to competitive systems.
I say this tongue in cheek but who can argue. Infect the world then sell them monitoring tools. Regarding software I see the cameras connect into either ivms4200 or Hik central has this and the metadata been tested at mass 16 cams or more coming into a central reporting and alerting repository as my experience is as soon as I bring multiple Hik sites into a central station it starts to give issues on metadata and corollation of sites and data never mind frame loss
It impacts what distance you can properly detect at. The sensor needs a certain number or pixels to accurately measure temperature. Exactly how many varies depending on who you talk to. I've heard things ranging from a 3x3 grid to "9 pixels across the region being measured", but no standard.
In the Hikvision models, range in the 160x120 is from ~3-5', and they spec the 384x288 for 4.5-9m.
There is a document which I've seen referenced by Hikvision with range information, but it's not easily located on their support sites. I'll post it when I locate it. There is range information in this video (starts around 2:24):
And in this one for some products, starting at ~0:49:
I hoped that this report would show advantages and disadvantages of such a system, but above all, it should focus on topics such as calibration, if there are standards that support the precision that manufacturers say their equipment has. What we have seen is an unrestrained run on these products without the least critical sense in their use leading to situations of false security and bad conceptions, at a time when prudence is necessary.
Update: Hikvision has now added claims of screening "up to 30 people" to their marketing.
However, the recommended measurement distance on this camera is either 0.8-1.5m or 1.5-3m, depending on lens, according to their marketing. Using the 6mm lens to reach 3m, the field of view width is only 1.3m wide (~4.26'), which would fit about 2 people across. So, unless you are attempting to measure subjects outside of Hikvision's own recommendations, 30 person measurement is not physically possible.
Those we've talked to in thermographic temperature measurement say that blackbodies should be recalibrated, every 6-12 months. They do drift over time. Exactly how much is going to vary depending on the blackbody. Like anything, quality will vary, some use better components than others.
That being said, more than one manufacturer has also told us that their blackbodies do not need to be recalibrated.
I would lean toward at the very least taking a temperature measurement of the blackbody on day 1 and making sure that six months later the temperature matches. I personally wouldn't leave it up to chance or manufacturer recommendations on something this critical.
A Hikvision salesperson told me that their thermal body temp cameras to not need blackbodys because they are more accurate than those from other manufacturers like Dahua that require them. I'd like to have comments from anyone who is technically knowledgable about these type of products as to whether this is a valid claim. If it is, it would put Hikvision in a very competitive posistion give the cost of blackbodys.
Is it accurate if set up in a 3’ wide area with a black body in a building where the person had been inside more than 5 minutes? Was that tested? Good luck making that happen in a public building but was it tested?
We do see Hikvision selling their 160x120 (thermal) resolution bullet cameras for as low as 1600-1800$ in Middle East and they claim that it works decent with .5 degree accuracy without black body.
At this price point and resolution is this achievable?
And for the 384x288 also they offer the same accuracy without black body and I am seeing clients buying these models in quantity of 100s.
Has IPVM done any testing using hik thermal cams with and without black body so far ?
Considering countries spending millions on these we expect a shootout from IPVM to be done between the atleast 2 models each from from top 5 prominent brands offering them.
This need to be done not just in closed environment but also considering that it’s places near entrance and the person coming in from a heated environment and from a very cool environment and how are these cameras handling it.
here lies the big challenge.In Middle East there are 10s or thousands of these sold already,I really wonder once summer kicks in what level of error is going to happen for people walking in from the sun at 45-50 degree Celsius and then forehead being scanned by these in a supermarket or mall entrance .
If IPVM could do a proper test this may save a lot of people 100s of thousands of dollars wasted because most of them still don’t know what they are buying for these kind of prices .
We have not yet tested Hikvision. We're trying to order it in the next day. Anyone with leads on a good place to purchase with stock or short lead time, email me at email@example.com, happy for any info. Most distributors show it as special order here which means anywhere from 7 days to some number of weeks, in our experience.
We are in the process of testing systems now. We released ourtest of Seek Scantoday. Any day, we will be receiving systems from Sunell and Dahua, along with a Uniview handheld temperature measurement camera, and ZKTeco and another access terminal with body temperature screening.
I agree with you on the concerns about screening in summer. Even in less drastic conditions, like the Eastern US, it still gets to near 40°C/over 100°F, and it will definitely warm up subjects' skin. Just 30 seconds in front of a radiator in our warehouse triggered a false fever detection in our test of Seek Scan:
This is likely to continue in other systems which use skin temperature to detect elevated temperature.
It should be stated that Hikvision only offers a measly 3-month warranty on its thermal temperature-monitoring solution. They should be ashamed of themselves. I wouldn't offer any brand new item to my customers if it has less than a 1-year warranty.
anything below 35C falls into the range of hypothermia...
if the effective detection range of this solution is <.3, then this example appears to be detecting the opposite of what they are claiming to screen against - and instead, identifying people with dangerously low 'body temperatures'.
obviously you are correct about the flawed placement and other screening parameters.