Did you know that in London there was one CCTV camera for every thirteen people at the end of 2020, with the total number of cameras reaching nearly 700,000?
Video cameras are everywhere. They have been used for decades in households, industries, public spaces, and intelligent police vehicles, so as such they are not a source of excitement. However, sometimes all it takes is to look at old things with new eyes and we can discover something new.
Digital cameras have taken a leap forward in recent years in terms of resolution, dynamic range, video capture speed, and light sensitivity. At the same time, the availability of storage has been improved, which, when combined with smart compression algorithms, allows for the recording and storing of videos on an unprecedented scale.
All this opens the door to completely new application possibilities based on image data analysis. Today, video capture analytics can be used to prevent shoplifting, prevent and detect accidents in factories, identify objects in logistics, and streamline and simplify police work.
The key to such applications are artificial intelligence (AI) and machine learning (ML) technologies, making it possible to analyse images and extract valuable intelligence. Figuratively speaking, software applications can “watch” video footage and thus replace human eyes.
The need for video analytics is particularly urgent in areas where the number of cameras used (and thus the amount of data recorded) is growing, as the number of cameras and video recordings that can be monitored or retrospectively analysed by humans is limited.
This is particularly limiting in law enforcement, where the usefulness and impact of cameras on work efficiency is determined by how quickly law enforcers can respond to recorded events rather than by the number of cameras they have available. According to some estimates, up to 98% of video footage will never be seen by human eyes. This is why the police usually do not respond to events recorded by cameras.
In countries such as the United Kingdom, the untapped potential of camera systems has long been noticed. Millions of cameras across the country generate petabytes of data. However, making the most out of this would require enormous human effort. This is why the local police deployed a smart video analysis system that can detect suspicious activity, track the movement of selected people, and detect faces.
With the progressive deployment of video cameras in police vehicles, it is natural to start thinking about the use of technologies able to analyse the recorded footage since it is often transmitted in real time to dispatch centres.
The key functionalities of today’s intelligent police vehicles are basically limited to recognizing licence plates (ANPR) or the type, model, and colour of vehicles, and recording the situation in front of and behind the police car by using cameras. Video recording helps eliminate potential corruption and provides evidence for the future investigation of incidents. However, it can also increase police officers’ security, provided that the footage is transmitted to a dispatch centre in real time and the dispatcher is physically able to simultaneously watch all screens and remain vigilant.
While real-time video transmission is no longer a technical problem with police departments taking the benefit of this option, dispatcher vigilance remains questionable and the ability of a single person to monitor multiple video streams will always be limited.
Unlike humans, a computer algorithm can “watch” a virtually unlimited number of video streams in real time, identify situations it is trained to notice, and send an alert about any particular event or suspicion. This can include physical skirmishes, collisions of vehicles or people, serious traffic violations, the detection of weapons, and even the automatic face detection of wanted or missing persons from camera footage.
In the future, smart video analysis may prove to be more than just a useful tool. Given the growing use of cameras in police vehicles and the natural limits of human vision and attention, such technologies may prove to be indispensable.
At Mosy, we are already working hard on complementing our camera systems with analytical capabilities and making them useful for the police as well as other components of the integrated rescue systems.