Overview

The Temperature Sensor is an Artificial Intelligence (AI) Facial Detection tool that uses infrared to detect temperature but only when a face is detected in its field of view (the blue oval). If guests and screening subjects frequently experience instances when the temperature scanner does not recognize their face or seem to be accurately reading temperatures, there are a few reasons why this can occur, and a few steps that can be taken to prevent or greatly mitigate these problems. 

Scanner Reading High/Low Temperatures

Move Temperature Scanner to a Place with Controlled Temperature

Since the Temperature Scanner is an external scanner, during very high heats, the body surface temperature of a person rises and doesn't have time to cool to their internal temperature quick enough. It's a challenge with external temperature devices such as a temperol scanner or an automated temperal scanner like you're using. The recommendation for a failed temperature check is to give the body time to acclimate to the ambient temperature outside of direct sunlight to differentiate between a fever and warm surface skin temperature from the heat of the sun or the high ambient temperature. Another option is to use fans or other cooling devices by or around the kiosk allowing the body to cool its surface temperature before doing a temp check.

Screening Settings

In 10.3.47 or later of the Kiosk, there is also an Advanced Screening Settings tab, where you can set the range of Temperatures you want the kiosk to consider valid. If the kiosk is given a temperature outside of this range, it will reject the reading and ask you to scan again.

https://support.savance.com/Knowledgebase/Article/View/751/0/how-to-change-screening-settings

Facial Recognition Not Detecting Someone

Too Much Light

An abundance of sunlight or very bright indoor lighting that prevents the temperature sensor from recognizing the contours of a person's face - this prevents the temperature read from being triggered.

Too Little Light/Too Much Shadow

Not enough light or too much shadow causing the same effect; a lack of contrast prevents the facial detection AI of the device from recognizing that a face is in the field of view.

Issues With Masks

Mask Detection Off But Mask On

When Mask Detection is set to No and the person is wearing a mask that takes up a considerable amount more of their face than a standard surgical mask, this produces the effect to avoid, namely, a lack of contrast preventing the AI's point matrix from detecting facial contours it can identify as a face. In this case, the person does not need to remove their mask. They only need to pull it down slightly so that it is below the bridge of their nose. With enough of their nose exposed, the eye sockets and nasal bridge present enough points of reference for the AI to recognize a face.

Mask Too Dark

Too dark of a mask in too dark of an environment. 

Mask Detection Off But Mask On

Too light of a mask in too light of an environment.

Lack of Contrast Between Face and Mask

 Lack of contrast between the color/shade of a mask and the color/shade of a person’s face can prevent facial detection. For example, someone with very pale skin and a white or off white mask or very dark skin and a darker shade of mask may have a problem. Lighting can help this situation although it is best to advise people to wear masks in colors that have a high contrast to their skin tone.

Addressing Issues with Masks

Setting Mask Detection to Yes can help to reduce the problems presented to the AI's detection matrix by masks. When this setting is enabled, the temperature sensor announces that a person is not wearing a mask if one is not detected. To ensure individuals' masks are detected, they should make sure the top of the mask rests above the bridge of the nose.

Normal Operation

Keeping the above concerns in mind when physically placing your Kiosks will help greatly with good performance of your screening solution. When the temperature sensing facial recognition device is positioned in an area with good ambient light and the bridge of the nose is exposed or covered as necessitated by the Mask Detection setting, we have continually seen reliably successful temperature reads for customers with hundreds of screenings a day.

Avoiding User Confusion

The questions created for visitors and staff to answer when screening are completely customizable. After creating the questions/prompts users will see, we recommend that several test runs be done after the Kiosk is placed. Based on the experience during tests, the temperature question should be edited to add applicable advice to educate users to the best practices for scanning at that particular location. For example, “Please Scan Your Temperature by looking at the temperature sensing camera, be sure to pull your mask down to below the bridge of your nose” or “Please Scan Your Temperature by looking at the temperature sensing camera. If you are wearing a mask and it is not recognized, make sure the top of it sits above the bridge of your nose”. The wording of this question will obviously depend on the environment where the individual kiosk(s) is placed. It is important to check with users about their experience for the first day or two the  solution is in place, so that their experience can be made as simple as possible using the customizable question prompts.