Beyond Security

Beyond Security

An evolution of video cameras will create new opportunities

Advances in AI are happening quickly. That is one of the reasons it is so exciting for some and potentially alarming for others. By focusing on AI, or more precisely, the subsets of AI known as machine and deep learning, results can fundamentally change the value proposition of physical security cameras and infrastructure for organizations large and small.

From the beginning of analog CCTV systems, video surveillance has been seen as a sunk cost and a part of doing business. It has been an expense as common as putting in windows and doors. Unlike windows and doors, modern AI-assisted network video cameras have steadily evolved beyond their original job of visually recording events to becoming smart sensors that can detect and describe objects and behaviors in increasing detail.

Moreover, modern AI-based video cameras are enabling operations and security departments to shift their stance from being purely reactive to past events, to proactively addressing situations in real time and preventing them from escalating.

AI vs. Analytics – What’s the difference?
In the video surveillance world, AI and analytics are often mentioned interchangeably, which can lead to confusion.

In the context of surveillance cameras, AI, or specifically deep learning, is used to visually detect objects within the camera’s field of view. An AI-based camera has been trained in an R&D lab to recognize a person or a vehicle, while also capturing a set of defining characteristics or attributes about the object. Characteristics might include the color of garments worn, approximate age and gender, and additional objects that might be worn or carried such as hats, glasses of bags.

Where analytics come in is in defining what the object is doing. Has it crossed a line or entered a zone? Is it loitering in one place beyond what we might expect? Maybe a car is travelling the wrong way down a one-way street? Counting people is another example of an analytic. These behaviors are then turned into data that travels with the video, making searching for objects with certain characteristics lightning fast compared to a human manually combing through hours of footage. If desired, analytics can also trigger alarms or notifications that can be sent to a security team.

To sum it up, in its most basic form, machine and deep learning algorithms (aka AI) detect and describe objects, while analytics analyze what the objects are doing and report on them.

Deep Learning and Extracting Data from Visual Imagery
When we talk about AI in the physical security industry, we are usually talking about a subset of the larger AI field. Machine learning and deep learning are both types of artificial intelligence that allow computers to learn without being explicitly programmed.

Machine learning is a broad term that encompasses any type of AI that learns from data, while deep learning is a subset of machine learning that uses artificial neural networks to learn from data. Artificial neural networks are inspired by the human brain, and they can be used to accomplish a wide variety of tasks, including image recognition, natural language processing (à la ChatGPT) and speech recognition.

One of the key advantages of deep learning is that it can learn from, and process through, large amounts of data. Video, compared to something like a spreadsheet of numbers, represents a lot of data. In the field of AI, video is often referred to as unstructured data (images, audio) versus structured data (numerical, categorical) that is more organized and easier to analyze.

Although unstructured data is much harder for deep learning algorithms to process, the algorithms can do amazingly well at image recognition when presented with enough training data. This is because artificial neural networks can learn complex patterns in data that would be difficult for traditional machine learning algorithms to learn.

It is for this reason that deep learning is particularly adept at identifying objects in images. Today, surveillance video has become a major contributor of unstructured “big” data. By using deep learning, we can make sense of the mountains of video that are being captured and have it help us find the proverbial needle in the haystack.

Doing More with Less
Deep learning has enabled security teams all over the world to be able to do a better job with the limited resources they have. Employing a surveillance camera with object recognition means that new and unique forensic searches can be done in a tiny fraction of the time it would take a team of people to do manually.

Likewise, because searches can be saved, we can be on the lookout for known vehicle types that are suspected of crimes in the area, for example. When that red van loiters outside the jewelry store at 4am, we want security teams to get an alert. If there is a person wandering around the loading docks after hours, security staff want to know about it. Deep learning is what enables security teams to evolve to a more proactive stance to potential threats, versus only reacting to past events.

Having a traditional “pixel-based” motion detection camera in the above scenarios would trigger an alert every time a shadow goes by, or a car headlight sweeps across the cameras field of view causing non-stop false alarms. Deep learning and object detection makes motion-based analytics truly useful.

Going Beyond Security
While the advantages of deep learning to the physical security market are easily understood, video surveillance cameras are also being used by organizations to analyze video footage and extract meaningful insights about sales and operations. By using deep learning algorithms to analyze video footage in real-time, businesses can extract customer behavior data such as foot traffic patterns (via heat maps) and product placement effectiveness.

By collecting and analyzing this data, organizations can improve their operations, optimize store layouts, and improve customer experiences. For example, retailers can use video analytics to identify popular shopping areas and adjust their store layouts to increase sales. They can also use data to optimize staffing levels, reduce queues, and improve inventory management.

In healthcare, video analytics can be used to monitor patient and visitor movement and improve the overall patient experience. In transportation, video analytics can be used to improve traffic flow and reduce accidents. And in manufacturing, video analytics can be used to monitor production lines and improve efficiency. The latest AI tools can memorize an entire scene and monitor stock on store shelves or in a warehouse and notify staff when inventory is running low.

Overall, AI-based video analytics are becoming increasingly important for organizations looking to harness the power of big data to improve their operations and make data-driven decisions.

Deep Learning, Image Enhancement
While deep learning is certainly a boon to analytics, these algorithms are also being used to enhance image quality and reduce network bandwidth. For example, deep learning can inform the image processing system in a camera as to which pixels represent a person or vehicle in motion. In this way, a camera can automatically reduce noise and ghosting in low-light conditions around a moving object without impacting static objects and the image background.

Similarly, knowing which pixels represent a known object can help the encoder prioritize those pixels over the background image. This improves encoding efficiency and saves valuable network bandwidth and storage without sacrificing quality.

As deep learning technology continues to develop quickly, we’re sure to see even more innovative and powerful applications for security, business, and operational intelligence.

The opportunities and applications unlocked by this impressive tool are limited only by our imagination. Any organization should be able to quickly identify ways in which this powerful technology can help stimulate growth and revenue while simultaneously protecting assets and individuals.