As part of a nationwide effort to prevent and contain COVID-19 transmissions, government buildings and public venues like shopping malls and workplaces have implemented temperature screening and access control measures. With Singapore gradually opening up, commercial buildings expecting high human traffic flows are using thermal cameras to identify and deny entry to visitors with high temperatures. However, this approach comes with certain difficulties. For example, conventional thermal cameras cannot adequately and accurately measure the temperatures of visitors when they put on headgear or have hairstyles with fringes. This is because thermal imagery is obscured when an individual’s forehead is blocked. Thus, requiring officers to intervene in order to obtain an accurate temperature reading. This is manually intensive and time-consuming, especially in places with high throughput. All these can be overcome with VigilantGantry, which allows for fully automated, contactless temperature screenings.
What is VigilantGantry?
Fig 1: VigilantGantry being used for temperature screening
VigilantGantry is a fully automated, contactless gantry system for temperature screening. It was developed by GovTech’s Data Science and Artificial Intelligence Division (DSAID). Driven by artificial intelligence (AI) and deep learning, it augments existing thermal scanners and gantry systems to improve the rate of contactless scanning, ease bottlenecks in long queues outside buildings and reduce the manpower required for temperature screening measures. In fact, the design thinking behind VigilantGantry was to leave no stone unturned. Hence, the term “Vigilant” was included in its name.
Why Should VigilantGantry be Adopted?
The key benefits of VigilantGantry are as shown below.
It is a well-rounded module system: Every component in VigilantGantry is modular, which means it can be programmed to adopt additional functionalities. For example, other than measuring body temperature and performing facial indexing, the system can also capture the location, date and time of screening sessions to support contact tracing purposes. It is also possible to store health and travel declaration data obtained via questionnaires.
Ease of scalability: VigilantGantry’s flexibility means that it can be scaled up easily. For example, it can be integrated into existing thermal systems without the need to purchase new equipment to support its integration. Given the shortage of thermal sensors amidst the COVID-19 pandemic, this is an additional bonus.
It is intelligent: Like a smart home hub, VigilantGantry links the gantry unit with existing thermal cameras and CCTV systems. Face recognition, forehead exposure and colour segmentation algorithms provide the intelligence required for proper screening. Once the thermal cameras detect high temperatures, a customisable alert will be triggered, and the visitor will be denied entry. Further checks can then be conducted on the visitor.
Fig 2: VigilantGantry checking for exposed skin on detected faces
Deep learning models will be used for face segmentation to quantify the amount of exposed skin on a detected face. When insufficient facial exposure is present and detected through the CCTV optical camera (e.g. if headgear reduces the available exposed skin for checks), VigilantGantry will send an alert to direct the visitor to take off their headgear before allowing entry.
Fig 3 VigilantGantry sends an alert in situations where there is insufficient facial exposure
Prevention of false alerts: With the YoloV3 person detection model, face segmentation is only triggered when a visitor is positioned within a pre-defined region of interest in the CCTV optical camera. This prevents the thermal camera from indiscriminately scanning and triggering false alerts from background heat signatures such as mobile phones.
How do You Use VigilantGantry?
GovTech has open-sourced VigilantGantry. System integrators may choose to adopt VigilantGantry by leveraging the open-source code
here.
Success Stories
A trial at the National University of Singapore (NUS) helped optimise the temperature screening process from one minute per visitor to an average of two seconds per visitor.
The team has since worked with Singapore-based companies who have indicated an interest to deploy VigilantGantry at high-traffic public spaces, so that entry screening operations are safe, efficient and accurate.
What's Next?
The
VigilantGantry source code
has since been open-sourced on GitHub. This will help expedite the adoption of this solution by more organisations in Singapore or abroad.
Contact Us
Reach out to the product team with your queries or feedback through this form.
Last updated 11 August 2022
Was this article useful?
Did this page help you? - Yes
Thanks for letting us know that this page is useful for you!
If you've got a moment, please tell us what we did right so that we
can do more of it.
Did this page help you? - No
Thanks for letting us know that this page still needs work to be
done.
If you've got a moment, please tell us how we can make this page
better.
