Less watching, more learning: AI enabling a smarter world

Presented by Western Digital

Video surveillance used to be about simple cameras continuously recording video and storing content day in and day out. Despite collecting seemingly endless cycles of video data, it would only be reviewed in the event of an incident. That is no longer the case.

With the emergence of AI-enabled applications, traditional surveillance is transforming into smart video with new use cases that transcend what we think of surveillance today. Industries from health care to transportation now leverage vast amounts of video data to deliver business value and improve quality of life.

Storage is the essential foundation to support enormous video data growth and feed AI analytics workloads. The worldwide video surveillance camera market is estimated to grow to $44 billion by 2025, up from $23.6 billion in 2019, according to IDC. Emerging smart video architectures that capture, store, and analyze massive amounts of video data and perform heavy deep-learning video analytics are putting an unprecedented level of stress on storage subsystems.

A “smarter” generation of AI-enabled use cases

Imagine a stadium with 60,000 seats. In the past, security personnel might monitor dozens of cameras to spot suspicious behavior.  Most of the time surveillance video was viewed only after an incident had occurred, and the remaining unseen video was eventually discarded.

Now think about how that same stadium can use smart video and AI to proactively plan better foot traffic or seating plans that account for appropriate social distancing. Can the flow of fans teach something about how to design or redesign stadiums or where to place the food courts or bathrooms? Is there a safety issue that can be prevented from causing injury?

Smart retailers use AI to understand their customers better. How many customers visit a particular aisle in a store?  How do customers react to different sections of the store? Did they stop by the seasonal display on aisle 10?

Smart city projects around the globe use data captured by traditional surveillance cameras to optimize traffic flow, re-envision safety by managing transportation fleets, and decrease pollution to improve quality of life.

Making smart video happen

AI-enabled smart video innovations use sophisticated algorithms to analyze video in real time and take actions that were previously left to humans.  Whether on the camera, NVR, or on the back-end server, deep learning analytics constantly improve these algorithms, requiring not only more storage capacity but also more sophisticated processors.  This results in true AI capabilities, such as searching for and identifying a pattern from weeks or months of stored video or creating traffic heat maps from hours of retail surveillance video.

For smart video to enable these new use cases and resulting insights, OEMs and integrators must design architectures that support AI workloads, with storage solutions that can handle multiple video streams per camera and feature high performance, capacity, workload capability, and reliability.

“Our customers rely on us for making their world a safer place with intelligent cloud-managed video surveillance and access control,” said Marton Renkis, VP OpenBlue Security and Innovation and GM Cloudvue at Johnson Controls. “We in turn rely on reliable and high-performing storage solutions to insure the integrity and security of our critical video surveillance and other physical security data.”

IDC expects there will be 41.6 billion connected IoT devices, or “things”, generating 79.4 zettabytes (ZB) of data in 2025, with video surveillance among the fastest growing segments. In addition to the sheer volume of data generated, the increased bandwidth required to store and stream 4K and beyond is growing. An AI application using 4K for pattern matching can increase accuracy significantly but requires more compute power and more storage.

Smart hospitals are managing multiple streams of data captured in 4K with frame rates of 45-60 frames per second. One example is the resolution required to visually see individual drips into a hospital patient’s I.V. As video resolutions and frame rates increase, the need for more storage capacity accelerates.

With new AI smart video workloads enabled by more connected smart security devices, it’s essential that the storage supporting them can sustain continuous read and write for long periods of time.  HDDs with a high workload rating will help reduce the amount of maintenance needed and will be able to withstand the constant write and erase cycles and many more hours compared to commercial and consumer-grade products.

Video and related metadata may also need to be retained longer on storage servers and system — on-premises or in the cloud — to analyze trends now or in the future. Therefore, storage reliability is a critical factor. The storage solution also needs to easily scale to meet growing demand as well as potential data retention policies and capacities required for various applications/use cases. For example, last fall a national sports league increased its data retention requirement for stadium footage from 14 days to 42 days. That’s a huge amount of storage that now needs to be considered.

AI-enabled smart video from endpoints to cloud

It’s no longer about just storing video, but what we can do with the video once captured.

Just as surveillance systems have evolved to smart video, so too have the storage solutions behind them. With the advancements in the capability of today’s AI-enabled smart video solutions, the amount of storage and compute resources required to capture, store, and analyze the video from the millions of cameras installed throughout the world is constantly growing. Every camera, recorder, and back-end analytics system will require more compute power and more robust storage to deliver on the promise of AI.

Western Digital recently announced the expansion of its WD Purple family with the WD Purple® Pro product line for the new generation of advanced AI-enabled cameras, recorders, and back-end servers. In collaboration with its ecosystems partners, the company is leading this shift to smart video by providing storage solutions that make smart video a reality.

Brian Mallari is Smart Video segment director at Western Digital.

Sponsored articles are content produced by a company that is either paying for the post or has a business relationship with VentureBeat, and they’re always clearly marked. Content produced by our editorial team is never influenced by advertisers or sponsors in any way. For more information, contact [email protected]

Source: Read Full Article