With camera counts in large-scale applications such as city surveillance becoming larger than ever, they generate huge amounts of data that strains human resources. According to a whitepaper by Quantum, this is where end users can benefit from video analytics, which is set to drive the future of video surveillance.
With camera counts in large-scale applications such as city surveillance becoming larger than ever, they generate huge amounts of data that strains human resources. This is where end users can benefit from intelligence and video analytics, which is set to drive the future of video surveillance.
That’s the argument presented by Quantum in its recent whitepaper titled “Intelligent Storage Enables Next-Generation Surveillance and Security Infrastructure.” It begins by laying out the challenges that end users are faced with today – a rising number of cameras that are higher in resolution and produce overwhelming amounts of data.
“While camera efficiency for preventing impunity is demonstrated and camera acceptance by the population is widely verified, huge deployments are underway in major cities around the world, with cumulated numbers of cameras for public transports, circulation and traffic control already exceeding 10,000. Technical infrastructure challenges are posed in terms of networks, processing power and storage space,” it said.
As a result, the end user can find help from video analytics, which has become more mature and proven effective in various applications. “This is clear with license plate recognition which is commonly used for automated speed enforcement or parking access control. It is also remarkable for face recognition, which was a complex dilemma that has been met today with an impressively high level of maturity. A broad range of applications use video analytics to complement the human operator in the supervision task,” it said.
According to the paper, analytics enhances video surveillance in two ways: video alarm triggers and assisted video search.
Video alarm triggers: In this scenario, the video analytic process is being applied to the real-time video stream and triggers an alarm when a situation is detected. The most simple of these video analytic filters include camera tampering, tail gating, counting, dwell and abandoned objects, among others. Those very simple algorithms filter only one video stream, and many camera manufacturers embed them directly on the IP camera board itself.
Assisted video search: Filters that prove efficient to detect certain types of situations may also be used to filter out sequences of interest in large video archives. The filter is processed by the NVR in charge of managing the recorded video feed.
Migration to predictive analytics
According to the whitepaper, video analytics that supports real-time detection of abnormal situations or helps search through video evidence during investigations is an absolute must, but the idea that such systems could even help avoid abnormal situations is compelling. “A large part of R&D is now dedicated to automating video analytics and leveraging their results with statistical tools that can actually predict the occurrence of specific situations,” it said. “This has a direct impact on the IT architecture of video surveillance systems that must take into account data extraction from video footage, data correlation, and storage. The need for an analytic-oriented middleware is pressing, together with a persistent data layer backed-up by a solid storage system.”
Product Adopted:Software