
In partnership with MOBOTIX
Industrial Temperature Monitoring
Scylla AI works on MOBOTIX thermographic cameras enabling continuous, real-time thermal scanning for lithium ion battery overheating, heating of transformers and electrical substations, skin surface temperature monitoring, and more.
Book a demo
Related materials

How AI Improves Physical Security in the Oil & Gas Industry
Maintaining the highest levels of safety and control at all times is critical for O&G facilities. Learn how AI helps enhance situational awareness and better protect their assets and workers.
Read more
How AI Video Analytics Help Video Monitoring Centers Save Money
Learn how implementing AI-powered analytics in video monitoring centers brings cost savings and a quick return on investment.
Read more
Legrand Mexico Facilities
Scylla Thermal Scanning AI technology was deployed at Legrand facilities in Mexico
Read moreFAQ
Scylla uses three primary accuracy-boosting algorithms:
a. Smart Targeting: Scylla’s computer vision core of Thermal Screening is utilized to estimate the regions of interest where thermal measurement needs to be performed. It targets only relevant objects – people's faces, which eliminates false measurements from irrelevant objects in the area. The major contribution to the measurement is provided by inner-canthus regions.
b. Smart measurement: Scylla Thermal Screening uses proprietary AI-powered algorithms to derive the final temperature value. It does not rely on single pixel value, nor on simple average/maximum value calculations, which proved to be less accurate in numerous experiments and tests. Most importantly, Scylla AI video analytics is powered through several measurements per person. To accurately achieve that, we track the person through the area of measurement and measure in each frame. The system then checks the acquainted values for statistical coherency before reporting. Thus the system has added intelligence to self-verify each value.
c. Auto self-calibration: Auto-calibration is another proprietary algorithm that dynamically derives the current average/norm of temperatures and detects the values which are above it.
The accurate measurement range is dependent on the camera specifications of choice. There is also a minimum distance you should limit the measurements. These parameters depend on a number of technical characteristics of the thermal camera, such as lens and resolution. Typical range examples: 1-3m, 3-7m, 4.5-9 m.
Scylla Thermal Scanning system works in four phases: Detection, Analysis, Identification, and Notification.
1. Detection: When a person walks in front of high-definition cameras, dozens of multiple images are captured simultaneously with conventional and thermal imaging technology, looking for elevated skin temperature, facial mask usage, and proper social distancing.
2. Analysis: Scylla Thermal Screening instantly processes all data received, looking for an elevated skin temperature of 38°C (100.4°F, configurable) or higher, proper facial mask usage, and social distancing protocols. When an abnormal temperature is detected, an alarm is activated providing local notification while simultaneously sending alerts to mobile devices for non-monitored areas.
3. Identification: Advanced facial recognition technology automatically scans the captured image, matching it against images of employees, visitors, or vendors found within its internal database.
4. Notification: When detection points are not monitored by personnel, system notifications are instantly sent to mobile devices or the notification method of choice. This allows security or designated staff to immediately identify, locate, mitigate non-compliant incidents, and/or isolate the identified person for additional thermal screening.
Scylla Thermal Screening is camera-agnostic. We have integrations with several thermal camera providers. As for requirements, we require a bi-spectrum camera (visual + thermal channel), thermal radiometric capability (as opposed to merely thermal imaging), and access to thermal raw data.
No. Scylla Thermal Screening is not meant for outdoor use. The reason is twofold: measuring hardware and subject of measurement.
All thermal camera providers explicitly restrict the exploitation with indoor controlled environments only.
Skin temperature can be greatly affected by environmental factors (sun, cold, etc.) and would not accurately reflect the real skin temperature when measured outdoors. In general, measurement sites should not be exposed to direct sunlight, strong drafts, and air-conditioning flows.
No. Thermal radiometry works by registering near-infrared radiation, which is being absorbed by glass and plastic. Thus the measured values will be affected by almost any media between the subject and camera, even a visually transparent one. Water and moisture also absorb IR radiation thus humidity of the air should also be considered.
Typically there is a limitation and recommendation of “no more than 95% humidity” when utilizing Thermal Screening.
Alerts can be received in three ways: Scylla web-based dashboard, Scylla mobile application, or e-mail notification.
Thermal Screening detects and alerts if a person being measured has a mask on or not. It also can provide monitoring of social distancing rules if applied.
Centralized solutions provide lower costs and simplified maintenance. Installation, setup, updates, and troubleshooting are in one place. Additionally, the servers are put in protected “war rooms”, in which the data will be safe, private, and secure rather than having many distributed installations that increase risk.
The waiting time will naturally depend on several conditions such as the difference of outdoor temperatures that the person was exposed to and the indoor temperature. For bigger adjustments, the wait is longer. The conditions that the person is waiting in also matter - if it is just a shelter with relatively high temperatures, the wait is longer compared to air conditioned cool indoor environments. These timings should be defined on-site with the hand-held temperature at particular outdoor conditions and followed.
Yes, we offer a cloud-hosted version. The decision between an on-premise installation and cloud-hosted installation will be based upon your budget. Feel free to request a free demo session and we can review the total cost of operation between both options.
If an individual is wearing a face mask, with an unobstructed view of their frontal forehead area, this will not affect obtaining a temperature reading. However, if an individual's complete facial area is covered with multilayered personal protection equipment, like a respirator, an accurate reading may not be possible.
The local network requirements have a few dependent metrics. For quick estimates we use the following calculation:
Number_of_the_cameras_in_the_network*bandwidth_of_visual_stream*2.
Typically the bandwidth is around 3-6 Mbs. If these cameras will be separately used for surveillance consider doubling the bandwidth. For rough estimates, 15 Mbps per camera is what we use for quick and dirty estimates.
LAN or WIFI connectivity: The majority of the thermal cameras have LAN connection only. Optional Wi-Fi connectivity is rarely offered. We strongly advise against using wireless networking as lower throughput causes issues not only for functionality but also in visualization parts of the solution.
For more details on network requirements, please contact Scylla Tech support.
Yes. All Scylla physical security solutions are GDPR and CCPA compliant. Scylla does not store any personal information and relies on real-time stream analysis.
Yes. Thermal Screening can be combined with any other Scylla solutions - object detection, perimeter intrusion detection, anomaly detection and behavior recognition, facial recognition, and many more security solutions. Feel free to request a free demo session. The additional Scylla solutions operate alongside Scylla Thermal Screening on the same server and even use the visual video feed of bi-spectrum cameras for analysis.
No. Scylla Thermal Scanning System supports both manned and unmanned scanning stations using alerts to send notifications to a dashboard and mobile apps.