Oh, the creative things you can think about with video analytics

John Merlino

When we talk about the potential of video analytics, I think Dr. Seuss actually expressed it best: “Oh, the THINKS you can think if only you try.” Let me explain.

While video analytics has long been a staple of the security industry, advances in technology and leaps in imagination are taking it well beyond traditional safety and security applications. Today you can find video analytics being applied to a host of operational tasks in business, education, health care and more. As users become more familiar with the technology, I’m finding they’re starting to use analytics in some surprisingly clever ways.

However, before we discuss how analytics are being used, let’s take a step back and think about what’s making all of this innovation possible.

THINK about image processing and storage

With each generation of video camera, we’ve seen substantial improvements in image resolution, frame rates, light sensitivity and even dynamic range. So we’re able to capture far more detail than ever before. At the same time, we’ve seen corresponding improvements in compression algorithms to reduce bandwidth consumption and storage. Video analytics both enhances and helps us bring structure and sense to that mountain of video data we collect so that we can quickly zero in on the most relevant footage. Video analytics can also be used to direct the cameras to record and send only video that is of interest, further reducing bandwidth and storage needs.

THINK about open application platforms

Video cameras that support an open platform are able to provide a framework for third-party programmers to develop and embed custom analytics directly on the camera. Additionally, the open platform enables multiple cameras to integrate with input from other technologies such as access control, environmental sensors, radar detectors and even network audio systems for a more holistic approach to monitoring, verifying and responding to events.

THINK about metadata and intelligence

It all comes down to metadata (the content information) generated by the camera. While video is unstructured data, video analytics can use the metadata to give context to the visual images and organize them in a way that makes them easy to find, understand and use. This has been invaluable in helping users quickly glean actionable intelligence from what they see. Essentially, video analytics turns cameras into smart observers – what many would call a processor with a lens. Now they’re able to detect, recognizing and classifying objects and ascertaining attributions like speed, direction, color and size. More sophisticated analytics can even distinguish demographics like age, sex and behavioral patterns. There are also analytics that use sentiment analysis to identify a person’s moods like happiness, sadness and anger. This ability to aggregate and parse video data in so many ways makes video analytics a super tool with endless possibilities.

THINK outside the box

Given the power and versatility that video analytics brings to security, it is little wonder why other businesses and those in local, state, and federal government and our military services are beginning to adapt the technology to address challenges in their own operations. For instance, many retailers and airports are using analytics like queue monitoring to reduce long wait lines and improve customer service. There are nightclubs using analytics like occupancy estimator to ensure crowd size doesn’t exceed legal limits. There are cities using directional analytics to spot wrong way drivers and vehicle counting analytics to control traffic lights dynamically. Stadiums are using advanced search analytics to find lost children and reunite them with their families. But these are all pretty conventional applications of video analytics.

The point of this article is to spark your imagination and help you think outside the box. So let me share a few scenarios you might not expect.

You’re probably familiar with people counter analytics; it counts the number of people passing by the camera. But one airport I know actually uses analytics to count people as they enter a restroom. Once a certain number of people have crossed the threshold, the camera automatically sends an alert to the cleaning crew to tidy up the room. It then resets itself for the next count.

A naval ship is using that same people counting analytic as a mustering tool to confirm that all crew members are on deck before they get underway. In engineering holds and areas that once required a manned watch or a rover, video sensors are being used to monitor gauges, validate operational conditions, and trigger alarms if negative conditions are detected so that they can be quickly remediated. With today’s ships operating with smaller crews, video analytics can provide sailors with the additional situational awareness they need to perform their duties.

In retail settings, demographic and sentiment analytics is not only being used to catch known shoplifters but also to trigger specific advertising messages on video displays based on the demographics (age and gender) of the shopper walking by.

Hospitals and assisted living centers are using video trip wire analytics to trigger a staff alert when a patient under medical restrictions attempts to wander off the premises or leave a locked down ward such as an Alzheimer’s unit. In nursing homes they’re integrating video analytics with acoustic analytics to better protect their vulnerable population. The acoustic analytics triggers an alert when certain changes in the ambient levels of noise are detected, such as aggressive voices, breaking glass or other anomalies. The video analytics residing in-camera verifies events and provides the situational context – the actionable data – that enables the facility manager to send the appropriate staff to respond.

THINK about deep learning and artificial intelligence (AI)

Today we’re starting to see a fusion between camera vision, which is no longer just a lens, and deep learning technologies which are enabling video analytics to become more accurate in extracting, classifying and cataloging metadata, smarter at tracking patterns and trending demographics, identifying hot spots and transforming video into usable intelligence. We’re seeing acoustic analytics learning over time about the environment in which they’re deployed so they’re better able to distinguish the difference between say a door slamming, a car backfiring or a gunshot; between a window breaking and a drinking glass shattering; between verbal aggression and exuberant conversation.

Feeding all that analytics information into AI engines not only provides us with an intelligent body of actionable data but serves as a basis for predicting future trends, patterns and behaviors, which helps us improve our own decision-making over time.

What sort of things will YOU think?

I hope reading about the different ways you can apply analytics becomes a springboard for your own ideas on how to achieve a greater return on your own analytics investment. I look forward to hearing about all the new thinks that YOU think and whatever your imagination might create.