alphaspirit - Fotolia

How AI in physical security makes public places safer

Deep learning-based tools are increasingly finding a home in physical security to enhance the protection of real-world assets and make public spaces safer.

As AI technologies like computer vision, pattern matching and predictive analytics continue to improve, businesses, governments and individuals are starting to apply AI in physical security environments.

The need to protect property and facilities is critical. Physical security equipment spans a wide range of applications, including electronically-enabled doors and gates, cameras, drones and even service animals meant to deter criminals. Video surveillance has been used for many decades at places such as airports, hotels, stores, malls and just about anywhere else people congregate or goods are sold. Each of these use cases is getting a boost from AI today.

Monitoring and access control of crowds

In areas where there are large gatherings, such as parades, festivals or marathons, or locations where many people need to pass through a security checkpoint, such as airports or sports stadiums, AI in physical security is helping keep a watchful eye over the crowd and detect and prevent threats. These places are increasingly making use of AI-enhanced image or video technology to allow crowds of people to pass through while still providing constant monitoring of both people and the surroundings.

Facial recognition technologies are starting to be adopted at airports to help with a variety of tasks to help decrease bottlenecks. For example, Dulles International Airport in the Washington, DC region also began using facial recognition boarding technology for certain international flights in 2018. The computer vision system is able to quickly scan a passenger's face and run it against either U.S. passport photos or a photo taken of foreign nationals when entering the U.S.

This new system helps streamline the once-manual process of airline employees checking every passenger boarding the flight. It also provides an additional layer of security as systems can scan for passengers flagged on no-fly lists or criminals trying to escape -- something human employees are not able to do.

Those responsible for safeguarding areas where large crowds gather are also turning to facial recognition technologies to help identity criminals. Notably, Chinese authorities are heavily invested in this technology. They were reportedly able to identify individual criminals from tens of thousands of fans in a stadium or in very crowded streets through the assistance of facial recognition technology. While currently undergoing much scrutiny, and in some locations, outright bans, facial recognition enabled by AI is becoming key to many modern security systems.

Smart surveillance systems

Local leaders in highly-populated cities are always looking for ways to stay on top of crime. As a result, surveillance in "smart cities" is being augmented by AI. AI can classify people, animals, vehicles and static objects to detect anomalies. Through examples, the system can determine what is considered normal and flag suspicious behavior for follow-up. Chinese company Hikvision is creating cameras with built-in deep neural networks for city streets, allowing these cameras to detect things like unattended bags or a suspicious car. For many reasons, it's not always possible to have police or security guards in these locations, so AI-enabled cameras help to keep a watchful eye all hours of the day.

In certain situations, because of poor video quality, bad lighting or camera angles, images from surveillance cameras can be low quality. AI can help to reconstitute images from blurry video surveillance cameras, allowing humans to get a better look at the footage.

Personal use of home video security systems is on the rise as well. High-end systems are starting to come equipped with AI features. These home security systems can automatically detect and flag suspicious behavior, alerting the homeowner only when something is out of the ordinary. As these systems become more widely adopted, it's only a matter of time before security systems like Ring come with a computer vision system installed.

Automation of repetitive tasks

Another area where AI in physical security is helping is through the automation of repetitive tasks. Traditional surveillance systems record hours and hours of footage but it's up to a human to review the tapes to spot suspicious activity. AI is now helping to distinguish between different images, eliminating the need for humans to do the laborious work of looking at hours of video footage. Instead now the AI system can classify various objects and people in images and determine pertinence. Motorola body-worn-cameras enable police officers to more efficiently search for objects or persons of interest.

As with all areas relating to safety and security, it's important not to fully take the human out of the loop. AI systems are great at flagging potential criminals, spotting anomalies and suspicious behavior, but they should be used as augmented tools to help humans do their jobs better rather than fully replace them. When used this way, they have the potential to be another set of eyes on large crowds, help keep lines and people moving and reduce crime or catch criminals. For these reasons, the use of AI-assisted physical security is sure to increase in the coming years.

Dig Deeper on AI hardware