The use of smart cameras for video surveillance brings a number of advantages for businesses, but also very specific challenges. But what requirements do these solutions have to fulfil in use cases to unlock their full potential? asks Manfred Berger, Senior Manager. Business Development Platforms EMEAI at Western Digital.
Artificial intelligence (AI) has long since found its way into almost all areas of our lives. Even in the business world, AI-based technologies are gaining a foothold. One field of application that holds considerable potential is intelligent cameras. More and more companies, including an increasing number of small businesses, are interested in smart video solutions, and for good reason.
Video surveillance is one of the fastest growing markets in data storage. Smart video is the evolution of traditional video surveillance, with the added power of AI and deep learning (DL). The cameras used in smart video can now analyse incoming video streams in real time. The insights extracted in the AI process allow decisions to be made without human intervention. In this way, smart video can act as the eyes and brain of a human, speeding tasks up through automation.
Unlimited application scenarios
The range of applications for smart video is huge. Uses range from traffic control in cities to alerts to warn people about natural disasters before they strike. Intelligent video processes can also be applied in retail and in industry. For example, in smart retail, the analysis of shoppers’ behaviours using motion data from cameras can help identify the best locations for product placement. In manufacturing, smart video can identify inefficient manufacturing processes or enable better quality controls. Even workplace health can be improved using the technology – for example, by monitoring the minimum distance required between people .
Offices and commercial buildings, along with schools, public buildings and retail locations have some of the greatest potential for implementing smart cameras . All kinds of businesses benefit from AI-supported security systems. In contrast to conventional surveillance solutions, AI in the camera or NVR help the business to filter out unnecessary data and allows for the comprehensive automatic analysis of objects as a number plate recognition. In addition, with AI, multiple live video streams on screens, or taped data can be observed and evaluated in parallel. The cameras and recorders can also be easily managed remotely.
Enormous challenges for technology
Despite the promise of today‘s AI-enabled implementations, smart video deployments also bring new challenges. To benefit from AI and smart video, businesses should define their requirements before choosing the technology.
Smart security systems typically consist of three main components: Camera, video recorder, and internal or cloud storage. In smart video solutions, AI comes into play either in the camera or on the network video recorder (NVR). The AI, in turn, requires DL that uses algorithms meant to function in a manner similar to the human brain. Video data can be processed on the NVR or, more commonly, on a DL server on the back end, or in the cloud. In either case, DL requires massive amounts of data to train the AI to enable useful analytics. Ultimately, the better a neural network is trained, the more accurately and safely the AI will function.
Effective AI also requires video data with the highest possible resolution, high frame rates and high levels of detail. 4K and 8K resolution is becoming increasingly prevalent in smart video solutions . This means that the amount of data to be processed is also becoming huge. The result is a need for ever larger and more reliable data storage. In addition, camera systems in companies or factories are often in use around the clock, 365 days a year. This continuous operation, with simultaneous recording and analysis of multiple parallel video streams, places enormous stress on the equipment and technology. If the camera is installed outdoors, the constant impact of wind and weather also plays a significant role. Downtime is not an option for smart video solutions if the cameras, streams and training process for the AI is to run efficiently.
To ensure seamless quality in capturing the video, businesses need reliable cameras, recorders and storage solutions. All the system’s components should be qualified and tested to be compatible with each other. They should be chosen for the specific needs of the business.
It is important to note that storage optimised for video surveillance differs from conventional hard disk drives. They are engineered to support multiple video streams, heavy “write” workloads and can feature firmware that improves streaming by reducing image loss and enabling high-definition (HD) recording. Storage solutions in the data centre must also be specifically designed for the flood of data generated by Deep Learning. The most important thing here is very high sequential performance when “reading” and “writing” data.
Smart video systems should therefore, first and foremost, be reliable, robust, and resilient overcome bottlenecks in storage capacity and performance.
In closing, companies have much to gain from the use of smart video solutions that can harness AI to bring new business insights. But, before rolling out such solutions, businesses should define their desired outcomes, and build a system to meet those requirements.
A security integration partner can help with this, and there are also online tools, such as calculators to determine surveillance storage needs based on the building size, the video quality and scene activity, that are helpful in this regard – eg Western Digital’s Surveillance Storage Capacity Estimator Tool. The list of factors to consider when selecting a smart video solution is long. However, the benefits of using smart video are just as numerous. – especially since AI is constantly evolving and creating new use cases.
