The Impact of COVID-19 and Masks on Video Surveillance Evidence

In today’s culture, reducing the spread of COVID-19 has become a priority in both public and professional spaces. The Centers for Disease Control and Prevention (CDC) recommends that the general population wear face masks while outside of their homes. This precautionary step mitigates the risk of viral spread between people. The requirement of a face mask, however, makes the identification of subjects through surveillance very difficult. And a person’s face is critical for identification in a majority of current surveillance techniques. Most video and image surveillance algorithms were not designed with face masks in mind since they were developed well before the coronavirus outbreak. Face masks also hinder interpersonal communication and limit our ability to read others’ intentions. As a result, surveillance technology upgrades have become necessary.

We previously discussed home video surveillance recommendations for doorstep theft during COVID-19. In this blog post, we will dive further into discussing effective COVID-19 surveillance and subject identification solutions specifically for those wearing face masks. Improved active and passive surveillance, clear masks, gait analysis, and exposure notifications are a few of the emerging solutions.

How Do Masks Affect Facial Recognition?

Most facial recognition algorithms created before the COVID-19 pandemic fare poorly when subjects wear masks. The National Institute of Standards and Technology (NIST) tested 89 commercial facial recognition algorithms. They found an error rate between 5% and 50% when matching photos of someone with or without a digital facemask. This one-to-one matching compares different photos of the same person, rather than one photo to a database, known as one-to-many matching. The study tested masks of various shapes, colors, and coverage and found a correlation between increased coverage and decreased matching accuracy. In addition, the mask’s color may play a role. Surgical blue masks caused less error than black masks. As NIST noted, none of the algorithms were created to accommodate masks, which is evident in the findings.

Do Masks Fool Facial Recognition?

The NIST study found an increase in false negatives but not false positives. In the case of a false negative, the algorithm cannot match two pictures of the same person, while a false positive matches photos of two different individuals. Masks mean these algorithms cannot reliably match different photos of the same person, but they do not falsely match two different people. These algorithms rely on all identifying features, including the eyes, nose, mouth, and facial bone structure, making facial recognition through a mask challenging. In some cases, algorithms could not process a masked face at all, resulting in failure to enroll (FTE) or template. FTE means the algorithm could not detect enough facial measurements from masked photos to begin making a comparison.

Coronavirus Security

Security during the pandemic has been a source of concern. The high margin of facial recognition error may cause traditional algorithms failure to hold up in court. As a result, improving facial recognition during the coronavirus pandemic is important. As long as the general population continues wearing facemasks, security must adapt. Learn about possible solutions to COVID-19 facial recognition issues.

Solutions to COVID-19 Surveillance and Subject Identification

Video Surveillance Upgrades

What if a closed-circuit television (CCTV) video surveillance system could identify a crime suspect without having to capture footage of their face? While these systems are widely utilized, many require a bump in quality in order to provide sufficient footage of a crime. Improvement in video recording quality is the key to keeping these CCTV video surveillance systems relevant, especially during COVID19 and the widespread use of face masks. Here are a few options for improving the quality of video recordings in both active and passive surveillance.

1. Active Surveillance

Advancements in technology coupled with rising crime rates have pushed businesses and law enforcement agencies to incorporate artificial intelligence within their video surveillance systems. These artificial intelligence solutions provide an additional set of eyes to help identify subjects, vehicles, and other details. Artificial intelligence combined with forward-looking infrared (FLIR) heat-sensitive cameras is extremely effective in the event of a crime or if a disguised subject exhibits symptoms of COVID-19, such as a fever. Infrared technology has been used for mass fever screening since the 2003 severe acute respiratory syndrome (SARS) outbreak. Fevers indicate many infectious diseases — the presence of a fever is a common factor to consider in healthcare screenings. Certain circumstances, including subjects’ biological sex, drug use, and the surrounding room temperature, can affect these screenings’ accuracy. Setting the right temperature cutoff is the most important element for avoiding false negatives without draining resources. Whether used for fever screening or facial recognition, FLIR cameras have proven useful.

FLIR cameras differ from thermal imaging in that they use short wavelength infrared light rather than mid-length or long wavelengths. They detect heat and turn it into an image. FLIR imaging is especially useful in low-light situations. To combat covert nighttime operations, United States Army researchers have implemented FLIR technology for better automatic and manual facial recognition. FLIR imaging is a possible solution for COVID-19 surveillance issues, as well as mass fever screening. In either case, this technology contributes to public safety efforts. Check out the video below, provided by Hikvision, to see how this technology works. Hikvision is a leading manufacturer of CCTV video surveillance technology.

2. Passive Surveillance

High-quality DVR systems, high-resolution cameras, and specialty cameras are just a few ways with which to improve the accuracy of passive surveillance systems. The highest resolution available at the consumer level is 4K. It provides the best possible picture for identification purposes in the investigation phase. License plate cameras and 360 cameras are two of the most common camera upgrades that are available for IP-based systems.

License Plate Surveillance used as Solutions to COVID19 Surveillance and Identification Solutions
360 Cameras

One of the most popular CCTV surveillance system upgrades is a 360 camera. The addition of a 360 camera allows for a larger focal viewing of a tight space. A 360 camera uses dual cameras to capture two images or videos, each with a 180-degree field of vision.

Passive surveillance 360 degree camera

The camera stitches the images together right away, or it comes with a companion software in which the user can stitch the images together after their capture. With a 360 camera, the odds of detecting a subject without a mask are higher. More complete footage lends itself to higher security. Check out a popular 360 camera here.

360 Cameras used as Solutions to COVID19 Surveillance and Subject Identification

Clear Masks

Clear masks address several issues regarding both security and interpersonal communication. Did you know that 55% of communication is visual? Traditional face masks, as we often see in the medical profession and in the recent months surrounding COVID-19, prevent our ability to see facial expressions and negatively affect our communication with others.

It can be difficult to read a person’s emotions when masks cover their nose and mouth. Humans prefer to look at faces as a whole when reading emotions rather than individual features. Traditional masks limit our ability to read middle and lower facial cues, which can be essential for expressing different emotions. For instance, a jaw-drop signals surprise, while a nose-wrinkle signals disgust. When people wear masks, they have to focus more on the eyes to read emotional signals, but prolonged eye-contact can seem aggressive and make the observed uncomfortable.

What’s more, masks also muffle sound and limit lip-reading capabilities. This is especially challenging for those who rely heavily on visual communication, such as the deaf and hard of hearing, children, and older adults. Popularizing clear face masks would eliminate these concerns while preserving public health.

In addition to the solutions to COVID-19 surveillance and subject identification we’ve discussed, ClearMask™ provides natural communication and eliminates barriers caused by traditional masks. Its transparency and comfort have been put to great use within the medical field, improving relationships between patients and staff.

Traditional Face Mask
ClearMask Aids as Solutions to COVID19 Surveillance and Subject Identification

Gait Analysis

Gait analysis is one possible solution to the coronavirus facial recognition issue. It’s a form of biometrics, which “is a technology that makes use of physiological or behavioral characteristics to authenticate or identify people [1]. The most commonly used biometric applications are fingerprint and iris-based identification.”

Although facial recognition using biometrics can be helpful in identifying suspects, it can be fooled. Disguised faces are common, especially now during COVID-19 and the requirement of facial masks. Learn more in the video below.

Gait analysis, on the other hand, is a form of biometrics that’s quite different from the rest. It is the measurement and interpretation of human movement patterns and kinetics. For example, surveillance footage could identify a subject by observing the way they walk or run. Gait recognition technology is still being developed, though a growing number of scientific journals explain the process. Below is an expert from one such journal:

“Human gait identification aims to identify people by a sequence of walking images. Comparing with fingerprint or iris-based identification, the most important advantage of gait identification is that it can be done at a distance. In this paper, silhouette correlation analysis based human identification approach is proposed. By background subtracting algorithm, the moving silhouette figure can be extracted from the walking images sequence. 

Every pixel in the silhouette has three dimensions: horizontal axis (x), vertical axis (y), and temporal axis (t). By moving every pixel in the silhouette image along these three dimensions, we can get a new silhouette. The correlation result between the original silhouette and the new one can be used as the raw feature of human gait. Discrete Fourier transform is used to extract features from this correlation result. Then, these features are normalized to minimize the affection of noise. Primary component analysis method reduces the features’ dimensions. Experiment based on CASIA database shows that this method has an encouraging recognition performance.”

Gait analysis technology is still in its early stages. Some of its current limitations include varying conditions. Under controlled circumstances, recognition fares well, but variables such as changes in the walking surfaces or the subject’s clothing can negatively influence the algorithm’s accuracy. These factors affect the way a person walks or runs. Walking uphill on a grassy surface is bound to look different from walking across flat pavement. Moving in heavy winter clothes may differ from moving in light summer clothes. Scientists are addressing such variables as they improve gait recognition programs.

The video below is an example of gait analysis.

Google and Apple's New Exposure Notification Technology

Google and Apple have joined forces to develop a technological solution for the mitigation of COVID-19 spread. The solution will improve the detection of COVID-19 outbreak hotspots and deter a subject from those areas, all on their handheld smartphone. Check out what NPR published on the topic below.

“The technology would rely on the Bluetooth signals that smartphones can both send out and receive. If a person tests positive for COVID-19, they could notify public health authorities through an app. Those public health apps would then alert anyone whose smartphones had come near the infected person’s phone in the prior 14 days. The technology would be available on both Google Android phones and Apple iPhones.

The companies insist that they will preserve smartphone users’ privacy. Smartphone users must opt-in to use it. The software will not collect data on users’ physical locations or their personally identifiable information. People who test positive would remain anonymous, both to the people who came in contact with them and to Apple and Google.”

The effectiveness of this technology depends on several factors. It will only help limit the spread of the virus if people are able and willing to participate. Everyone, or almost everyone, must have access to both a smartphone and quick-response COVID-19 testing. They must have opted into the exposure notification settings and be willing to notify others of a positive test result. Not everyone has access to a smartphone, and not everyone has access to reliable, quick-response COVID-19 testing. Many feel wary of location-tracking on their smartphones and turn off these features when not in use. These individuals may prioritize their privacy and choose not to share positive results, despite the companies’ promises not to collect identifiable information. Even if all the above criteria are met, a notification does not guarantee action. Knowledge of exposure might not keep someone from entering the public and infecting others.

Professionals in surveillance and cybersecurity point out these concerns. Jennifer Granick, ACLU surveillance and cybersecurity counsel, responded, “No contact tracing app can be fully effective until there is widespread, free, and quick testing and equitable access to healthcare. These systems also can’t be effective if people don’t trust them.

Although this “Exposure Notification” may be useful in the future, it also may be useful in the world of video surveillance. However, several larger steps must come first.

  1. The public must learn to trust the technology.
  2. The technology must pass all civil liberties questions in order to minimize invasions of privacy, abuse, and stigmatization.
  3. The technology must integrate into surveillance technology, such as CCTV video surveillance systems.

To learn more about the “Exposure Notification” technology, follow this link. Despite the current pitfalls, exposure notifications may become more helpful if they meet privacy standards and increase in popularity. It’s only logical to apply advanced modern technology in the fight against COVID-19.

Contact an Expert

For more information on the solutions to COVID-19 surveillance and subject identification discussed here, call or email us today. We’re experts in audio, video, and image forensic services. We help decipher surveillance content, providing understandable, court-friendly reports. We also aid in forensic services and evidence acquisition, can serve as an expert witness, and more.

This article was last updated on February 22, 2021.