DATA ANALYTICS - FYMA

The video surveillance revolution: from passive to ultra-smart video data analysis By Karen K. Burns, CEO, Fyma.

Over the last few decades, we’ve seen a huge shift in the way security and CCTV cameras function. From the outdated and unsophisticated video surveillance systems that passively recorded still images and grainy video footage, to modern ‘smart’ cameras that can differentiate between still and moving objects, people’s faces, car number plates and much more.

There are over 700 million CCTV cameras in the world, just less than half of those are estimated to be located outside of China. This leaves us with 340 million cameras – each day, every single camera streams massive amounts of data, leading to petabytes of it being gathered. Yet the most important data – how people and objects behave and interact – is largely left uncaptured.

AI vision – a new dawn in video analytics

Computer vision AI can now enable video cameras to analyse and collect powerfully specific data insights such as differentiating between cars, motorcycles and scooters, as well as quantifying biometric traits such as height, gender and movement patterns of subjects. What's more, the AI technology can be trained to analyse whatever metrics users want to quantify, provided that the AI solution is fed and trained the right information in an ethical and legal way.

This is incredibly important as vision AI solutions should be built with privacy and security front of mind in order to safeguard easily identifiable sensitive personal data.

A successful approach for achieving this is teaching the vision AI solution to only pick up and analyse human data subjects from the neck downwards, thus never seeing their faces. This means that when the AI technology is used by cameras in real life, people’s faces are never detected or fed back into the respective platform or system whilst still being able to track movements across targeted areas.

When it comes to analysing biometric data that is subject to regional data protection laws, vision AI solutions should be able to achieve 100% compliance at all times.

From a commercial standpoint, video surveillance solutions leveraging AI technologies can open up a whole new world of opportunities for organisations. They will be able to garner endless streams of data insights about consumer behaviours in a way that is both ethical and GDPR compliant

Earlier this year, Dutch Railway company NS deployed computer vision enabled drones to manage crowds and flows of traffic between local trains stations and the Zandvoort F1 circuit during the Dutch Formula 1 Grand Prix.

The drones, embedded with computer vision AI, helped to track and analyse footfall of the thousands of spectators at the event in real time in order to ensure operational capacity of trains whilst allowing for social distancing measures. Such deployment of drones helped provide better journey safety and security for passengers, along with a better all-around customer experience.

It’s clear to see that video surveillance systems powered by vision AI are the natural successors to the much-maligned one-dimensional video monitoring that many organisations still rely upon today. AI vision will enable video monitoring teams to map and respond to consumer behavioural patterns in real-time, and by extension predict future behaviours for better decision making.

Whilst traditional CCTV surveillance cameras do the basic job of passively capturing and recording events, video surveillance is no longer about retrospectively reviewing past events, it’s about gaining deeper insights into the behaviours of people, objects, and vehicles in real-time.

Given the ubiquity of traditional CCTV cameras, organisations utilising them could more than benefit from implementing new sensor technologies such as computer vision AI into their existing security and CCTV camera operations to help derive better data insights of their surroundings. This includes retailers that can gain better analytics for capturing and analysing shopper behaviour to build more accurate customer profiles, city councils that can integrate the tech into their vast public CCTV systems to secure better safety and security of the public, as well as traffic management systems which can use vision AI for more advanced monitoring of pedestrians, cyclists and vehicles in and around urban areas.

Organisations need to ensure that they are not left behind. By moving away from the old, inefficient and rigid way of recording and managing visual data, they can instead start adopting smarter, sophisticated, and more data driven video surveillance solutions that can unlock new revenue streams and improve existing operations.


Simon Spring, Operations Director of EMEA at WhereScape, discusses some of the common misconceptions surrounding data warehouses, data lakes, and data hubs, and why a proper understanding of each is key to building effective data infrastructure.
By Davide Villa, Director of Business Development EMEAI, Western Digital.
By Bhushan Patil, Chief Growth Officer for Network Services for APJI & EMEA,
By Sharon Einstein, General Manager, Customer Engagement Analytics, NICE.
By Jonathan Westley, Chief Data Officer, Experian UK&I.
The success of environmental policymaking is inextricably linked to the deployment of high-frequency data and artificial intelligence (AI). Geoff McGrath, Managing Director of CKDelta, discusses the role of holistic, data-driven decision making in responding to the climate crisis and driving down carbon emissions.
By Aaron Regis, Senior Solutions Engineer, TigerGraph EMEA.
What does the Edge mean for IoT? Peter Ruffley, CEO, Zizo, explains how edge IoT and analytics can provide a powerful mechanism for translating complex data sources into a streamlined, lower cost platform with faster return on investment and higher value. First, he considers some of the key challenges businesses face when considering an investment in IoT.