How to manage the video data tsunami with intelligent platforms

Systems designers are regularly challenged by the exponential expansion of advanced sensor technology and how to implement them into a security or life/safety environment. A network of sensors can hear, smell and feel events like audio alerts, gas leaks and temperature fluctuations. However, it is video surveillance that has emerged as the dominant force in security as a result of emerging intelligent cloud-based analytics that can tame and harness enormous stores of disparate data enabling actionable and real-time response.

In a recent report, – The Physical Security Business 2020 to 2025,  global research firm Memoori stated: “that in 2019, artificial intelligence (AI) technology applied to video surveillance convinced the market that by 2020 it will become mainstream over the next 10 years and that there is a critical need to make full use of the massive amounts of data being generated by video surveillance cameras and AI-based solutions are the only practical answer. Modern chip architecture with AI software can comb through vast volumes of data and boost security and safety. Granted, there is a lot of development in this field that we are yet to see, but the path towards AI seems quite clear.”

The Memoori report added that “AI adds a brain to our smart technology ecosystems, which allows these systems to understand the wide range of sensory inputs and process massive amounts of data being provided by sensor hardware. This is especially important for computer vision that feeds from numerous streams of video surveillance footage and creates far more data than would be feasible for human teams to monitor. Like the human brain, AI video analytics can count people, identify individuals, and recognize activities in order to drive security, efficiency, and behavioral prediction applications, and even help us fight crime.”

The Migration from VMS to AI

The simple fact is that the security industry is making a transition from traditional VMS into a sensor management platform for data collecting. When you talk to end-users and technologists alike, migration is inevitable. The stress of data collection, data generation and eventual data interpretation is becoming more complex each day. An enormous amount of data is created every day, roughly 80% of that is unstructured or ‘dark data’, while 37% of that is video data. This creates a heavy burden on people to locate, organize and input data before the process even begins.

The challenge of turning dark data into real-time actionable intelligence at the edge is the mission Fredrik Wallberg and the technology team at Airship AI have embarked on. Wallberg, the Vice President of Marketing, says that Airship has looked out at the landscape of data and security and what's coming next noting that their three-letter agencies and corporate client are creating a ton of data but still struggle with tangible information gathering.

“We (security) have moved away from live viewing almost all together. We've seen plenty of reports of the monkey that doesn't get seen by people that are watching it live because they are simply overwhelmed. So, a lot of the videos are being watched in playback mode,” says Wallberg, noting this important trend away from live monitoring and are exhibiting at ISC West 2022 this week in Las Vegas “What we've done through existing customers and some big Fortune 100-plus-type companies that we work with is bring a deep expertise in ingesting existing data and metadata and extracting that from the video source to further increase security and also enable business.”

Wallberg explains that Airship AI’s ability to create a process that helps clients get real-time data visualization at the edge is a key differentiator. “Organizations do a lot of the computing and a lot of the data generation at the edge. Our Airship outpost AI box is really just a small-form factory and video edge processing module that provides onboard AI data analysis and transmission. With these ever-expanding sensors and IoT devices that are being connected to the internet, we'll probably see this as a trend moving forward. We feel we're at the leading edge of it when it comes to managing the data points and the workload at the point of data creation for maximum efficiency and accuracy.”

What To Do With All That Data

After the Airship AI process has regenerated and collected all this data, the next step is the analysis of the data. Wallberg says that's when you pull it back into the core management platform so the data can be ingested at the edge to organize and analyze, it turns all that dark data into some real-time actionable intelligence.

“At this point, much like a lot of the competitors, you do need that powerful backend with at least a simplified user interface (UI) on the front end to ensure that the data silos are knocked down and replaced with data hubs. The AI engine can then see all the structure or light data -- if we call that as the opposite of dark data -- in the core,” adds Wallberg. “From an AI standpoint, you are creating all this data and you're making sense of that data by aligning and connecting the backend system, fusing and operationalizing the data where visualization becomes reality. The last piece is what we've labeled as data visualization in this sort of data thread. And that's really just in providing intelligence through the entire data flow using AI. You’ve taken all these disparate systems, knocking down these data silos, taking the IoT devices and sensors at the edge, ingesting the data, analyzing it through the core AI functionality, getting that real-time data visualization at the edge, and then generating real-time notifications. So, while you have a super smart engine at the edge and a powerful engine at the core and back end, at the end of the day, it needs to be simplified.”