With artificial intelligence starting to have an impact on how we go about our daily lives, it would only be proper for us to relay how it is currently affecting CCTV systems. As CCTV installers with over 20 years of experience in the industry, we have seen much change in that time – the prospect of AI has added to the excitement of the vast improvements in security cameras in recent years.
Here, we are going to go through these improvements before asking how the latest of these, AI, will affect security and building operations going forward.
How Recent Improvements In Tech Are Currently Affecting CCTV
Internet Protocol (IP) cameras capture images digitally and are able to send and receive data via a computer network and the internet.Unlike 'traditional' CCTV (that capture images in analogue and has closed circuits),IP systems are able to transmit pictures in high-resolution which can be viewed remotely from any type of internet-connecting device.
With video camera analytics, a user will have the option to automatically analyse video to detect and determine any behaviours or incidents of interest. These functionalities take different forms; one of the more simpler being motion detection against a fixed background scene. The basis of how CCTV works in technology is formed by camera analytics.
Utilising those analytical software programs, AI is able to scan images to recognise humans, product displays, vehicles and their positions, as well as their behaviours. Aside from 'simple rules' – such as restricting access to humans and vehicles at certain times of the day, more'complex rules' can be set. This may include if the user wants to track the direction of how a human is moving, or whether to monitor any vehicles that enter the premises, rather than those who are leaving.
- Rule-based intelligence refers to systems that have been programmed by CCTV installers with set tasks.
- Active intelligence sees the AI learn what is 'normal behaviour' in the environment it is designed to monitor, alerting the user if it deems something to be unnatural.
How We're Currently Using AI
Most of the AI software used today is focused on being rule-based. Currently, AI is used to increased the abilities of camera detection and to improve its efficiency.
Improving The Capabilities of Cameras
One example of rule-based AI in action, is its loiter detection capabilities. This is based on predetermined rules set up by CCTV installers that will trigger an alarm if any individual loiters for longer than the predetermined period of time. Another example is hotspot detection; using a thermal analytics camera,it can detect the temperature, which if it rises above the predetermined point,will trigger an alarm.
The ability of AI to gather, recognise and learn from data is well known – which can only help to cut down on the number of false alarms that are triggered. In everyday use, the operational capability of a camera maybe compromised by environmental aspects – the weather and insects being the main culprits. Insects in particular are known for restricting the view of cameras, sometimes even triggering a false alarm – with machine learning, AI can be fed information of what a certain insect is and instead of triggering an alarm, it will instead send a message to the operator that the camera needs cleaning.
How Will AI CCTV Develop?
Like its machine learning abilities, the development of artificial intelligence in security cameras never stops. Here are some of the future developments of AI CCTV:
Quick Searching (With Natural Language)
Imagine if you could just type certain behaviours into a console and it would present incidents that match that description? This is possible with AI – and it will only improve as time goes on. By using tagged metadata and deep algorithms, AI software can recover clips based on what the user enters into the search. As an example, if they wanted to find occurrences of 'dropped box', it would use what it has been programmed/learned to recognise what a 'drop' and what a 'box' is, and search its footage of occurrences it has marked of it happening. This is known as 'reactive' – searching out incidents after they've happened. Currently in development are 'proactive' systems that would detect incidents before they happen, based on pre-programmed and learned behaviours by CCTV installers.
Using cameras to analyse footfall statistics are already providing retailers with the means to run their businesses better. Whilst these cameras are generally installed over entrances to stores, what if a larger space needs to monitored? Not only can they cover a larger area better but footfall analytics by AI systems can prove useful for a wider range of behaviours too – such as how long a person dwells over certain displays, how they react to them and which displays are the most-visited. This information allows retailers to better understand the effectiveness of their store, optimising their layouts as a result.
Counting People, Age and Genders
Whilst existing cameras are adept at calculating footfall,they aren't able to provide a comprehensive list of statistics than what AI can. Its software is able to use face recognition to achieve a wide range of stats – including the ages and genders of visitors. For public buildings especially, they can use images of people (whether they're banned from the building, known offenders or suspected of doing something) and send alerts whenever the camera picks them out.
So that is where AI CCTV is at now and how it will develop in the near-future. If you are interested in using it to improve the operation of your business, why not get in touch with us to learn more?