Technological breakthroughs and the adoption of digital technologies have brought great progress, and it can sometimes be hard for the untrained eye to recognize how these cutting-edge innovations, such as those being deployed in smart cities, are visibly improving safety, convenience, and the standard of living for us all.
Smart cities use visual information and communication technologies to improve operational efficiencies, share information with the public and provide a better quality of local authority services. For example, advancements in Internet of Things (IoT) technologies have enabled connected public transportation systems, which leverage real-time monitoring capabilities, as well as tracking the locations and routes of public vehicles. Not only does this speed up service times and reduce traffic congestion, but it also cuts waiting times for passengers and keeps them informed of delays or emergencies.
There are security and safety aspects to smart cities as well. ‘Smart’ video or cameras utilize AI algorithms and deep learning (DL) to analyze visual data in real-time and can dispatch orders from a hub to AI-powered devices much faster than a human can process. And beyond just providing data, smart technologies can actually enable the devices to deploy intelligence and insights. For example, cameras and AI-analyzed traffic patterns can adjust traffic lights accordingly to improve vehicle flow, reduce congestion and pollution, and, more importantly, increase pedestrian safety.
Smart video is also being deployed in smart cities to deliver critical assistance to help recognize and curb crime. Business owners, for example, find security cameras helpful in protecting their property, reducing shoplifting, and monitoring for unusual activity or employee or customer incidents. On a larger scale, real-time video analysis can identify and differentiate between objects, such as distinguishing humans from animals, and issuing an alert when they are in a prohibited location.
Cameras Get Smart
Smart cameras need to ‘learn’ to recognize objects and events and sort the identified activities into categories such as anomalous or normal. This is where AI and DL are needed for training and learning. DL must analyze a huge amount of data to identify patterns and achieve accuracy. The development of higher video resolutions, such as 4K, is key here, enabling CCTV cameras to capture more data in higher quality and from various angles, which makes the analysis easier.
The smart video market is going through a transitional period – one that must enable recording video at scale. This means shifting away from simply recording raw visual data on a standard camera, to carrying out analysis on the AI-enabled camera itself. In the past, it was only possible to conduct the data analysis at a centralized location, such as a data center, but today the rise of on-board AI chips allows this analytical load to be distributed. The ability to distribute the work is crucial when working at the scale of a smart city, enabling the data to be processed more quickly at the endpoints.
As AI and 4K rise in adoption on smart video cameras, higher video resolutions are driving the demand for more data to be stored on-camera. And with the many types of cameras being used today, such as body cameras, dashboard cameras and new Internet of Things (IoT) devices and sensors—more storage is needed to analyze data and pull insights from it in real-time, instead of post-event.
As a result, storage is critical to the evolution and efficient working of smart video systems. Smart video architectures require innovative storage technologies, which deliver needed flexibility, performance, capacity and reliability. Robust on-board storage must be specially designed to meet the needs arising from multi-streaming devices, on-device deep learning systems and AI-training solutions.
Storage solutions for cameras and recorders have evolved to provide high data transfer and write speeds, as well as the capacity to ensure world-class video capture.
Where Storage Meets AI
Having improved workload and performance is important in ensuring that drives can keep up with the demands of AI functionality, including pattern matching and object recognition. By combining video stream recording optimization with top-tier durability and capacity, smart video solutions and AI-analytics have the necessary foundations in place to operate at optimum levels for thousands of hours.
NVRs and VMS (video management systems) are getting smarter. Deep-learning algorithms go beyond tracking simple motion detection, to enabling advanced capabilities to drive improvements in many industries and settings, including retail, smart cities, and entertainment, to name a few. AI-enabled VMS are being architected for new graphic processing units (GPU) and central processing units (CPU) to improve overall deep-learning capability, and speed algorithms related to object identification.
NVRs with this deep learning require greater storage capacity and more sophisticated processing, versus individual cameras, enabling them to perform more advanced analytics, such as finding a particular image from weeks or months of stored video, or creation of traffic heat maps from hours of retail surveillance video.
Choosing the Right Architecture
It is clear that smart video plays a vital role in increasing public safety such as monitoring and adjusting streetlights and promoting safe driving practices in fleets of vehicles, to detecting flaws and deviations in a product line to avoid potentially dangerous products from moving through an assembly line. Home security systems such as smart protection and alert systems also benefit from the role of smart video.
However, the success of smart video relies upon a robust and resilient storage architecture that can effectively keep up with the heavy workloads created by multiple camera streams, and related AI and analysis of these workloads. As video use cases rise and grow across industries and smart cities, the requirements of data storage can’t be an afterthought.
About the Author:
Brian Mallari is Director of Business Segment Marketing, HDD Business Unit at Western Digital. As part of his role, he manages Western Digital’s WD Purple smart video storage products. Learn more here.
