In the physical security space, video analytics have historically over-promised and under-delivered, often leaving end users sceptical about their capabilities. However, increased integration with security solutions and other business systems, as well as developments in deep learning and artificial intelligence (AI), have given video analytics a significant boost in recent years. Here, we take a look at the key trends putting video analytics in the spotlight, and how this opens up new opportunities for increased security and business intelligence.
At the start of 2018, security industry experts commented on how deep learning technology and Artificial Intelligence (AI) would extend to the video surveillance industry, allowing security professionals to gain very specific insights into human behaviour. Our experts predicted that this would permit organisations to reduce risk, enable efficiencies, reduce costs, ensure compliance and provide faster access to stored video. With AI-enables video systems, video analytics are set to perform more complex applications at a higher level of accuracy.
According to Ambarella’s Chris Day, advancing chip technology combined with the neural network approach to computer vision is game changing for video analytics. Since the problem of higher resolution has already been solved, the key differentiator for video surveillance systems will be the ability to add computer vision in parallel with image processing and high-resolution encoding – ideally in a chip that is low-power.
Video systems produce an immense amount of data that is often wasted, says Bosch Security Systems’ Sean Murphy. When video analytics alerts are integrated with other security systems, video events can trigger responses from other parts of the security solution. For example, cameras with video analytics can initiate intrusion detection system events initiate intrusion detection system events, prompting the panel to take action by alerting the central station or sending video to security personnel.
Adding network video to the current generation of Internet of Things (IoT) solutions provides actional value beyond situational intelligence for security purposes. With increasingly intelligent sensors, interactions between business systems are becoming more sophisticated, providing a value greater than the sum of the parts. Organisations can use smart applications to reduce energy consumption, allocate workspace, and reduce operating costs. In a retail environment, analytics are now capable of assessing a scene for occupancy and crowd control, even generating reports of trends over time.
Camera-based video analytics can go beyond assessing a current scene to predicting potential risks before they occur, explains Pelco’s Jonathan Lewitt. Based on predetermined factors or analysis of prior events, systems can collect all available information to determine the level of severity of a situation and whether an action needs to be taken. At the same time, systems can correlate data from video and other sources to help analyse similar occurrences in the future.
Audio analytics are often overlooked, notes Hanwha Techwin’s Paul Kong, perhaps due to differing privacy laws from video surveillance. However, audio analytics processed in a camera can help provide a secondary layer of verification for events, as well as identifying gunshots, screams, or other sounds indicating an incident is taking place. This makes audio analytics ideal for dealing with active shooter events at schools and campuses. As Louroe Electronics’ Richard Brent explains, audio analytics software can detect rising levels of human aggression, as well as recognising firearm discharge. This can trigger alerts to ensure incidents are dealt with swiftly
