Sightcorp logo

Face Mask Detection

Comply with C-19 regulations

✔Check individuals and crowds wearing masks in public

✔Use digital screens to remind visitors to wear masks

✔Alert staff when no masks are detected

✔Works with existing USB or IP cameras with RTSP streams

✔Anonymous & spoof proof

Everything you need to get started with face mask detection


✔Real-time analysis

✔Crowds or individuals

✔Spoof proof detection

✔Out-of-the-box technology

✔USB / IP / RTSP camera

✔Linux & Windows

✔Easily turn on and off

✔Anonymous analysis

Use Cases


Retailers need to monitor their premises to control the current occupancy and wearing of masks. Digital screens can be used to display information for both the number of people allowed in the store and mask detection. Learn more about our occupancy monitoring system here

Public transport

Wearing face masks in public transport will be mandatory in many parts of the world. Public transport organizations can use the software to automate the checking process with very little resources needed.

Corporate buildings

With many office buildings opening up and employees coming back to work, face mask detection can be used to maintain a safe environment for everyone. The mask detection system can also be combined with facial recognition which uses employee database information to match an individual at the office entrance.


Our face mask detector can be very effectively used at airports mainly for entrance flow management and monitoring. The software can be added to any access gate or entrance to make sure that all passengers follow the safety rules when boarding a plane.


Hotels, restaurants and bars are opening their doors to the public with certain regulations. In many cases, visitors will be required to wear masks when checking in or interacting with the staff


Because cars usually have very limited space especially when travelling with multiple people, there is no option to keep the standard 1,5m distance. Thus enforcing masks is important for minimizing the threat of spreading infection


The wearing of face masks is especially important in hospitals and healthcare facilities. Our face mask system can monitor staff and patients to see if they are wearing masks during their stay in the hospital.

How does facemask detection work?

Face mask detection is an AI based technology that analyzes a video stream to detect and recognize a face mask worn by an individual person or a crowd of people. Our DeepSight software outputs a confidence value for each detection. Every individual is classified either as ‘wearing a mask’ or flagged as ‘not wearing a mask’. If the face mask detector application identifies a user as not wearing a mask, a custom message can be delivered via a digital screen to remind all visitors to wear masks before entering the premises.

How to wear a mask correctly?

According to the guidelines set by the World Health Organization (WHO) a face mask needs to cover the face fully, including the nose and the chin. Therefore our detector only classifies someone as wearing a mask if these conditions are satisfied. Our software is also spoof proof which means that it understands if you’re covering your face with a hand or an object other than a mask.

See how face mask detection works

Face mask detection deep learning

Our Face Mask Detection System is based on the latest deep learning algorithms built in-house by our own team of skilled Artificial Intelligence Scientists. Our system is very simple to use and it is built on top of Deepsight Toolkit.  The solution runs on the edge and can be connected to any existing or new IP cameras to detect people without a mask. The live output can be shared with any CMS to display custom messages on digital screens to alert store visitors to put a mask on or it can be connected to access gates to deny people access if they are not wearing a mask for instance at airports. This allows for the system to function fully automatically to enforce the wearing of the mask without any need for human intervention.

Face mask detection Camera

Our Face Mask Detection System can work with existing USB or IP cameras and CCTV cameras combined. It makes use of the latest Computer Vision algorithms to detect people with and without masks. The accuracy of our lightweight model is very high and it can detect if you’re wearing a mask correctly or not (ie.; mask covering the nose). Face mask detection data can be used for various purposes and industries for instance at corporate offices, retail stores, public transportation, airports and taxis.

Audience analytics and face mask detection

Several countries around the world have made the wearing of masks compulsory within the public domain such as shopping malls, public transport and grocery stores. Companies active in Retail, Digital Signage and DOOH industries often use our software for counting the number of people entering their stores and measuring impression and viewer numbers for their digital displays and advertising screens. Therefore it’s very important that our technology remains robust and accurate even when analyzing faces covered by face masks. Sightcorp has recently augmented their software with face mask detection. The face mask detection feature is an out-of-the-box technology for the analysis of faces partially occluded by masks. Older technologies based on machine learning would most likely ignore a large portion of the population with a mask but our models have been developed and re-trained with this limitation in mind and are still able to detect and count, determine the gender and estimate the age of all audiences. All of these models run in parallel so retailers can collect demographic data and make sure their visitors wear face masks during shopping at the same time.

Face mask detection software

Sightcorp’s face mask detection software is built on top of their existing audience measurement software DeepSight. The software offers real-time analysis of USB or IP cameras using RTSP streams. It runs fully offline and locally on both Linux & Windows platforms. DeepSight with face mask detection offers anonymized face mask detection which is completely privacy-proof: no user personal data is retrieved or stored as the system anonymously checks facemasks only. This technology can be used on crowds or individuals and it’s easy to disable once face masks are no longer mandatory. With face mask detection disabled, DeepSight can analyze and collect useful audience data around viewing behaviour, demographics and mood. Learn more about DeepSight here.


How does face mask detection work?

Our face mask detection technology uses a combination of the latest face detection and face analysis computer vision algorithms to detect and analyze faces with and without masks.

Does it work in real-time?

Yes, our face mask detection can work with existing USB, IP and CCTV cameras in real-time.

What is the accuracy?

Our mask detection model performs with a 98% accuracy rate (tested on a reduced MAFA dataset).

Is face mask detection anonymous?

Yes, face mask detection offers anonymized analysis which is completely privacy-proof: no user personal data is retrieved or stored as the system anonymously checks facemasks only.

Can I customize my own confidence thresholds?

Yes, the software outputs a confidence value for each detection in the raw data output. Users are able to set their own thresholds according to their specific scenario. The software also outputs a yes/no value for each ID in the aggregated data output which uses a predefined threshold.

Can I try face mask detection?

Yes, we offer 2-week free trials for all of our products. Contact us for more details.

How is the data stored?

DeepSight with face mask detection is running locally on the edge device and all data is processed and stored on the same device by default. The software does not store any personal data, only information about face masks.

Do I need to develop or is it ready to use?

Face masks detection is a part of our existing DeepSight products which are ready-to-use solutions.

What hardware do I need?

Face mask detection has the same minimum requirements as our DeepSight products. Find more information about preferred devices here

Ready to try face mask detection?

Technical Specifications

The table below shows how FaceMatch SDK performs on the Labelled Faces in the Wild (LFW) dataset:

FPRTPRThreshold (Inverse of distance)
0.10.99900 ±0.002130.55448
0.010.99667 ±0.005370.59791
0.0010.99367 ±0.006050.62989

FPR = False Positive Rate
TPR = True Positive Rate

These results are an indication only and are based on the specific dataset Labelled Faces in the Wild. Customers can expect similar performance, with possible variations due to hardware and the availability of annotated data.