Video Analytics: Accuracy is everything!

Uri Guterman, Head of Product & Marketing for Hanwha Techwin Europe, looks at the development of video analytics.

It was not long ago that video analytics had a poor reputation within the electronic security sector. As an emerging technology it was overhyped, with users being disappointed with the accuracy of early software. Fast forward to 2020 and COVID-19, video analytics running on board open platform IP network video surveillance cameras is proving its value by helping businesses to reopen, as well as giving their colleagues and customers confidence that the need for a safe distance, is respected.

Manufacturers such as Hanwha Techwin have introduced a range of COVID-19 related applications, already pre-loaded onto cameras and are ready to work out-of-the-box.

Face Mask Detection and Occupancy

Wearing a mask prevents the spread of COVID-19 and has been adopted as a safety measure. However, the availability of the Face Mask Detection application could not come at a better time with the World Health Organisation recommending face masks when social distancing is difficult. Using video analytics to detect if someone is not wearing a mask, the application will trigger the playing of a audio message such as ‘please wear a face mask’.

An Alarm Out feature can be used to turn on a device such as a warning beacon. The application, which analyses up to 4 people up to 5 metres from a camera, is not affected by glasses, hats or scarfs, and is able to detect if a mask is being worn incorrectly. Pubs and clubs, will be desperate to reopen at the earliest opportunity, but in order to do so, might have to employ additional door staff to count the number of people who enter and leave. For companies already struggling, the cost of hiring these staff could be prohibitive. Designed to be installed at entrances, the application provides data (with accuracy) on the number of customers in a premises.

It does so by counting people entering or leaving a building. The application also generates alarm outputs to control automatic doors, as well as ‘wait’ or ‘enter’ messages.

Social Distancing and crime prevention

This utilises AI deep learning video analytics to help implement social distancing with accuracy. It can measure the distance between people in a camera’s field of view and generates an alarm if social distancing rules is not being complied with.

The Social Distance Measuring application performs equally well outdoors, offering an effective social distancing implementation tool for public transport environments and in city centres. Video analytics is increasingly playing an important role in crime prevention.

Many video surveillance cameras are now equipped as standard with suites of Intelligent Video Analytics, including enter/exit direction, tripwire and face detection. Combinations of these can be configured to form a cost-effective digital barrier.

AI Cameras

The deep learning video analytics detects various object types, including people, vehicles, faces and license plates, supported by AI algorithms which identify the attributes of objects or people; age group or gender. The algorithms are able to recognise if a person is wearing glasses or holding a bag. Operators are able to take advantage of VMS from leading software developers to forensically search for specific objects or people, by filtering the recorded meta data. This ability to minimise time wasting and costly false alarms means control room operators and security personnel are able to focus on responding to real incidents and emergencies.

ANPR

The introduction of serverless ANPR applications has made the technology affordable as a crime prevention tool for business’ or organisations. They are 95% accurate, and designed to offer a solution for car parks, petrol stations and small residential estates with multiple entrances/exits, barriers or roadways.

The ANPR camera provides the opportunity to automatically control the movement of 1,000 white listed cars through barriers via camera relay outputs, as well as provide car park management information, such as ‘time spent’ and occupancy rates, without having the cost of installing and running the application on a server.

Video analytics during 2020 and beyond

Watch out for even greater accuracy in video analytics technology which within the not too distant future are likely to enhance video surveillance systems to a level which at this time might not seem possible. Many of the innovations on the horizon will give even greater assistance to businesses as they seek to protect their assets and property, whilst creating a safe environment for their colleagues, customers and visitors.

No more articles