Temperature Screening Update – 9/20
We wanted to update our information on thermal temperature screening. The best source for that is IPVM which tests all types of devices and reports on the “bad actors”. We do not like having to comment negatively on events but when it is in the interest of the public and American taxpayers we have to speak up.
No Blackbody Mistake, Half Million Dollar, Hikvision Fever Camera System in Georgia
By: Charles Rollet, Published on Sep 16, 2020 | Email This
A Georgia school district touted buying Hikvision fever screening “about $2,000 cheaper per camera” with no blackbody for a total cost of a half a million dollars, despite the fact that these calibration devices are crucial for accuracy.
Moreover, the School district made a mistake confusing the accuracy of the Hikvision system with a blackbody to the worse performance of the ones they bought without a blackbody.
This news follows an Alabama school district spending over $1 million on a similar no blackbody Hikvision fever camera system and even declared it would not follow FDA guidance. Such deals harm public health efforts and waste taxpayer money while helping the bottom line of Hikvision and its dealers.
Georgia Deal Background
On August 10, the Fayette County Public Schools unanimously approved the purchase of “a maximum of 75” cameras for $7,000 each for a “maximum of $525,000” total, as shown in a video uploaded by the district:
The district ended up buying 70 cameras for $490,000, and the system is currently deployed at all of its schools, it confirmed to IPVM.
No Blackbody, Integrator “Did Not Recommend”
The district told IPVM that no blackbodies were included explaining that “the vendor [Ultimate Security] did not recommend that we purchase the blackbody.”
“About $2,000 Cheaper A Camera”
During the board meeting, Assistant Superintendent Mike Sanders touted Hikvision’s lower cost, saying he knew of another school system paying $2,000 more per camera:
I’ll just say I do know of another school system – they are currently installing these – I won’t mention the school system name because I do know we’re buying them, installed, about $2,000 cheaper a camera than what they are, they’re getting the same exact camera, we feel like it’s a good price point for us, we’ve seen it from $2,000 a device up to $100,000 per device, the prices are coming down now that more and more people are getting into the market [emphasis added]
Mistaken Accuracy, Confusing With and Without Blackbody
While the school district thought they were saving money, they actually confused the specified accuracy of the Hikvision system with a blackbody vs what they actually bought, without it. During the meeting Assistant Principal Sanders touts the accuracy of what they are buying as ±0.5°F, saying:
scrolls on down to the best and this, the long range camera, does fall into that best it’s an accuracy of point five degrees Fahrenheit and which is is is pretty good.
To the contrary, the Hikvision camera, sans blackbody, they bought only has a specified accuracy of ±0.9°F, 80% worse than with a blackbody. In this configuration, a reading of 99.5°F might be as low as 98.6°F or as high as 100.4°F.
The video below shows the school district claim and the mistake made:
Blackbody Crucial to Accuracy, IPVM Testing Shows
A blackbody (pictured right) is a calibration device that emits a single stable temperature, giving thermal cameras a crucial comparison tool. Below is an FDA diagram of “proper thermal imaging room setup”, note the blackbody adjacent to the person being screened:
IPVM testing of Hikvision’s fever cams shows that deploying them without a blackbody results in reduced accuracy/missed elevated temps/false alerts, even worse than the inferior accuracy Hikvision specifies.
IPVM’s test shows that without a blackbody, accuracy is typically reduced to within ±1.0°F to ±1.5°F, significant enough to result in false alarms and missed elevated temps:
The lack of a blackbody also increases long term accuracy risk as ambient temperatures change.
No Operators Touted, Despite FDA Guidance
During the August 10 meeting, Assistant Superintendent Sanders stated “another benefit” of Hikvision’s system is that it could be run remotely (no operator on the ground), with the system automatically taking pictures of high temp students and sending these directly to the principal, who would then “locate that student” to be sent to the nurse:
another benefit of this one is that we don’t have to have an administrator standing right there. The others, you had have to have someone standing there making sure they came in and stopped. With this one, it can be set to where an alert comes through to the principal, to the assistant principal, with a photo of that individual telling what the temperature read, and then that would allow them if someone’s not at the door, that would allow them the opportunity to go and locate that student [emphasis added]
However, the FDA’s fever screening guidelines instruct that each thermal imaging system be run by a “properly” trained person who prepares “the person being evaluated” in order increase accuracy:
Furthermore, running a fever camera system remotely means potentially contagious children cannot be immediately screened out and sent to the nurse, leaving them free to roam around and contaminate other students.
Quick Screenings Touted, Despite Inaccuracy Risk
Superintendent Dr Jody Barrow touted Hikvision screening “large numbers of students in a quick fashion”. Assistant Superintendent Mike Sanders also said the Hikvision system allows people to be screened without stopping:
most everything we had demoed would require a student to come up and physically stop, take the scan, and then move on, and obviously at the entry point when you need that, that’s not real feasible
However, in practice, if individual students are not made to stop and position themselves correctly in front of the camera and remove face/head coverings, the system will be much more inaccurate. This increases the risk of simultaneous screening, which the FDA does not permit, instructing users to “measure only one person’s surface skin temperature at a time”.
Debunked Hot Water Test Touted
As proof that the Hikvision system ‘works’, Superintendent Barrow noted Hikvision demos of people holding hot water cups getting “caught”:
some of their video clips, where they were actually running a lot of people through, they would take hot cups of water or coffee so they would make sure they caught that person
IPVM’s testing and explanation show how hot water bottle demos fool the public:
District Gets Forehead Right
One aspect the district got right is that Hikvision fever cameras only screen foreheads, so signage will be deployed to tell students to remove their hats, with Superintendent Barrow noting that “students aren’t going to be able to wear their hats as they come in”. On the other hand, since there will be no operator, students could easily ignore this. Moreover, hair bangs also interfere with Hikvision’s forehead only system.
Controversy Over Privacy, Human Rights
The Hikvision system sparked controversy locally; at an August 24 district meeting, one man in attendance criticized it over “privacy issues”, per local paper The Citizen:
In response to the privacy concerns, the school district says facial recognition is not part of the deployment, stating “the cameras are being used to solely check body temperatures, and not for any other identifying purpose.” The district did not directly address Hikvision’s human rights record after we asked.
Fayette School System Response
The district sent IPVM a response to our request for comment. In the response, the district quoted Ultimate Security president Billy Griffith defending the Hikvision choice, stating:
I personally contacted several locations in my research, in addition to four supply distribution centers, to make sure that Hikvision still had the reputation that I am familiar with, and my research confirmed it
Hikvision has aggressively promoted fever camera sales worldwide but has done little to ensure these are properly deployed as required by the FDA and international standards. The installation of expensive but poorly-implemented systems of questionable use is an inevitable result of this campaign.
IPVM Devices Tests
PVM has tested, by far, the most “fever” / temperature screening devices in the world. The following is a listing of our tests to date:
- Aratek “Fever Detection” Terminal Tested Yet (Aratek)
- Avigilon Elevated Temperature Detection Camera Tested
- Axis Thermal Camera / Grekkom Human Temperature Control Tested
- Dahua Thermal Temperature Monitoring System Tested
- Dahua Temperature Measurement Station Tested
- Dotty “Hot Or Not” Elevated Body Temperature App Tested
- FLIR A Series Temperature Screening Cameras Tested
- FLIR Screen-EST Screening Software Tested
- Hikvision and Uniview Entry Level Thermal Handheld Cameras Tested
- Hikvision Temperature Screening Tested
- Hikvision MinMoe Temperature Screening Terminal Tested
- Injes Tiny Temperature Terminal Tested
- Kentix SmartXcan Fever Screening Device Tested
- Mobotix Thermal Body Temperature Detection Examined
- K3 Wall Mounted IR Temperature Gun Tested
- K3 Pro Wall Mounted IR Gun Tested
- K7 Wall Mounted IR Temp Gun Tested
- Seek Scan Thermal Temperature Screening System Tested
- Smartphone Thermal Camera Body Temperature Measurement Tested (FLIR / Seek)
- Sperry West / Alibaba Tablet Temperature Measurement Tested
- Sunell Panda Cam Body Temperature Measurement Camera Tested
- ZKTeco Body Temperature and Mask Detection Reader Tested
- TVT Temperature Measurement Terminal Tested
- Uniview Wrist Temperature Reader Tested
Plus general tests:
Axis Compares Fever Camera Sellers to 9/11
By: John Honovich, Published on Sep 18, 2020 | Email This
PUBLIC – This article does not require an IPVM membership. Feel free to share.
Axis Communications, the West’s largest surveillance camera manufacturer, has quietly sat out the fever camera gold rush. No more.
In a new NBC News investigation into fever cameras in schools, Axis spoke out against this, comparing them to opportunistic sellers in the aftermath of the terror attacks of Sept. 11, 2001.
However, what fever camera salesmen are doing is much worse. While post 9/11 sales were very bad and many people bought useless or broken equipment, many fever cameras are rigged giving a false sense of security that increases the risk that more die.
Axis To NBC News
Below is what Axis told NBC News:
Some of the more established vendors of security products to schools, including Axis Communications, have avoided the fever detection business altogether over concerns that thermal cameras aren’t accurate enough to detect fevers quickly in large groups because of how sunshine, exertion and masks can alter readings.
“If you have 500 students entering between 8:00 a.m. and 8:15 a.m., we don’t believe we are going to make a product work accurately enough,” said Fredrik Nilsson, vice president of the Americas at Axis Communications, a major supplier of surveillance cameras to schools, retailers and other businesses. “We are very long term and not opportunistic.”
“If you have 500 students entering between 8:00 a.m. and 8:15 a.m., we don’t believe we are going to make a product work accurately enough.”
He compared the proliferation of companies selling fever detection tools to those that sold facial recognition technology in the aftermath of the terror attacks of Sept. 11, 2001.
“A lot of airports went out and spent hundreds of thousands of dollars to install facial recognition at check-in and realized it wasn’t accurate enough and they had tens of thousands of false positives a day,” he said, warning that schools could find themselves in a similar predicament. [emphasis added]
Axis certainly has a point here. But our testing shows it is even worse.
While undoubtedly hundreds of millions were rashly spent after 9/11 on video surveillance systems that were ineffective, it was much easier to see ineffective video analytics or poor quality cameras than it is to see ineffective fever detection.
When a perimeter protection systems alarms on a headlight, the user can immediately see the system is ‘broken’. When a facial recognition system alarms on someone who looks nothing like the watchlist, the user can immediately see the system is ‘broken’.
When a fever detection system misses someone with a fever, people cannot simply look at a person and tell (as they can with video analytics and facial recognition mistakes). Worse, since so few people have a fever and almost everyone nearly has the same temperature, it is easy to trick people into thinking it is working.
This has created the phenomenon of rigged systems:
It is good that Axis spoke out on this issue and we encourage more industry professionals to do so.