5 Ways AI-enabled Video is Transforming the Surveillance Industry
Artificial Intelligence, or more accurately machine learning and deep learning, is on the verge of transforming every aspect of our lives. Each time that you ask Siri for the weather, get Netflix recommendations on what to watch next, or unlock your phone with your face, you’re benefiting from artificial intelligence (AI). The power of AI is undeniable, but nowhere is that power more evident than in the video surveillance industry.
Wait! Why is a pure-play cloud storage company talking about AI and video surveillance? Well, for one, many of our customers use Wasabi hot cloud storage to archive their video surveillance footage. Our breakthrough price-performance enables them to store more data longer. They can dramatically increase retention periods while reducing the need to expand costly and complex on-prem storage arrays. Second, Wasabi partners, such as AXIS Communications, are doing some pretty amazing stuff with AI-enabled cameras and object detection and analysis.
They’re not alone. The business directory Crunchbase lists more than 900 companies that are applying AI to computer vision and eight that are doing so specifically for physical security. Allied Market Research expects the global AI-enabled video surveillance market to grow at a 15% compound annual rate to $55.2 billion by 2030, up from $14.8 billion in 2020.
Advancements in related technologies are converging to revolutionize video surveillance, including improved video compression and high-resolution cameras that can observe and analyze a wider-than-ever field of view. Artificial intelligence algorithms can then be applied to spot patterns, faces, and objects that enhance the quality and value of that surveillance. This pays off not only in better security but also as an improved customer experience. Here are five ways AI, coupled with these other advances, is revolutionizing video surveillance.
- Improved alert accuracy
Conventional motion detection systems have always been vulnerable to false alerts. They have difficulty distinguishing between a child riding by on a bicycle or rustling leaves during a windstorm and a legitimate security threat. The result is that security staff become so inundated with these alarms that they eventually stop paying attention entirely. AI-equipped surveillance systems can not only distinguish between people, objects, and vehicles at a fine level of granularity but also detect actions that warrant suspicion, such as the same person walking repeatedly back and forth in front of an exit door. The result is that AI-enhanced alerts are more meaningful, which makes it easier for guards and responders to concentrate on the ones that matter.
- Better object recognition
The 2013 Boston Marathon bombing was caused by two booby-trapped backpacks left on a sidewalk. Unattended bags have long been regarded as a security threat but they aren’t easily picked up by motion detectors. AI-equipped surveillance not only enables objects to be identified with a high degree of precision but also classified according to factors like shape, size, and the likelihood that they belong to nearby people. Smart cameras can even look for suspicious bulges or protrusions that indicate that an otherwise ordinary object has been modified or contains dangerous items.
- Crowd dynamics
An estimated 2,400 people died in the holy city of Mecca in Saudi Arabia when they were crushed and trampled by a crowd surge. Unfortunately, such incidents are not uncommon. One study estimated that more than 230 people died and more than 66,000 were injured in crowd-crush incidents between 1992 and 2002. AI enables the real-time movement of crowds to be matched against known patterns and anomalies in order to detect potentially dangerous situations. For example, if a large group of people is running in the same direction they may be attempting to escape a threat such as an active shooter. Conversely, large throngs of people gathering in a common area may forewarn the potential formation of an angry mob. Crowd dynamics has special relevance in the age of COVID because social distancing is likely to be an issue for some time to come. AI-enabled surveillance can measure crowd density and alert event organizers and building managers when certain thresholds are breached. People who aren’t wearing masks in violation of health protocols can be spotted and asked to comply. Inside the office, staggered work schedules can be maintained with the help of facial recognition that verifies that employees are where they are supposed to be. There are also more common uses, such as identifying traffic jams before they become gridlock or detecting long lines forming at concession stands and alerting event organizers to dispatch more resources to serve waiting customers.
- Enhanced authentication
The quality of facial recognition software is now so high that machines can achieve accuracy scores of up to 99.97%, according to the National Institute of Standards and Technology. Facial recognition software is now good enough to act as a powerful companion to conventional access control systems such as key cards and passcodes, which are vulnerable to being stolen or spoofed. Seamless authentication can also improve the customer experience. VIPs visiting a company can be whisked through security without delay and presented with customized messaging like greetings and directions to a meeting.
- Improved customer experiences
Video surveillance is no longer the purview of security alone. Cameras mounted in retail stores can enhance customer experience by improving traffic flows, identifying bottlenecks, and even streamlining the shopping experience. For example, Smart Retail solutions now include checkout-free stores with no cashiers, no self-checkout hassles, and no waiting. Instead, cameras and image recognition software identify each of the products in a shopper’s cart, enabling customers to simply walk out of the store with their purchased items. AI-enabled video can spot anomalies, such as a person wandering back and forth in apparent confusion, and dispatch an employee to help. Or it can identify an important shopper as she enters the store and alert staff to roll out the red carpet.
Forensics could be considered a sixth area where AI is transforming the video surveillance industry. While AI is certainly a boon to law enforcement for enhancing investigations, there is substantial public concern that AI-enhanced video or images could misinform authorities or be used as improper evidence in court. In fact, much has been written about the potential misuse of AI. The risks to privacy and personal liberty shouldn’t be underestimated. But neither should the immense benefits to businesses and society at large, which is why just about every major enterprise and government entity exploring the use of AI has embarked on a Responsible AI initiative to ensure fairness, transparency, and a lack of bias in their AI systems.
Another innovation converging on surveillance: Next-generation cloud storage
At Wasabi, we treat all data equally with one inexpensive, ultra-fast tier of service. Our highly parallelized system architecture delivers breakthrough price performance, enabling our customers to store 5x more data for the same price than with competing first-generation cloud storage vendors. This makes Wasabi an ideal storage target for video surveillance systems. And with our Object Lock immutable storage feature, no one can delete, alter, or tamper with files that could potentially become evidence, assuring a chain of custody that is critical in law enforcement or other CJIS-compliant use cases.
With the dark side of AI, such as deep fakes and other computer vision techniques for altering images on the rise, affordable storage immutability is no longer a nice to have, but a must-have capability in your video arsenal.