Intelligent video surveillance: an untapped source of value
Today in France, a growing number of public (streets, roads, parks) and private (car parks, airports, shops or cinemas) spaces are equipped with video surveillance cameras in order to protect the population and prevent possible crimes. Indeed, a great effort has been made to install cameras on the streets. The latest estimates for 2012 indicate the presence of 935,000 cameras in France, most of which are used to monitor companies and stores. For public places, the figure is estimated at about 70,000(1).
Today, these cameras are underused and lack the staff to read their information. Indeed, it often happens that there is no one to view the images, or that only one person is responsible for viewing multiple video surveillance screens at the same time, or that the analysis of video data is done after the event in the case of an accident or a crime. It therefore seems that surveillance cameras today are deployed only for deterrence purposes and simply serve to reassure the public. Thus, various reviews of video surveillance policies are critical of the multiplication of cameras as they are considered costly with very few results.
This being said, video surveillance has a real potential whose effectiveness could be greatly improved by Artificial Intelligence which, thanks to visual recognition, could finally enable it to keep its promises. Indeed, visual recognition is a part of AI that allows information to be analyzed and extracted automatically from an image.
This growing technology makes it possible to create high value-added use cases by detecting or predicting actions or elements of interest in a video. It would therefore be a pity for public authorities and companies not to take advantage of this advanced technology, especially since intelligent video surveillance represents only a marginal additional cost because the camera parks are already installed and functional.
Here are some examples to illustrate how visual recognition will increase cameras so that they can finally play their role of protection and detection to improve life in society. In its initial function, security, intelligent video surveillance can detect a threat in real time or identify unusual and suspicious human behaviour, send an alert automatically and thus trigger the intended reaction (human intervention, dissuasive voice message or other). But it is also useful, via the same set of cameras, in other safety situations or for business cases specific to each company.
In the field of safety, visual recognition makes it possible, for example, to detect a person who is unwell, to predict a potential danger such as a collision on public roads, or to detect the failure to wear safety equipment on construction sites. As for business issues, in the field of product marketing for example, visual recognition can identify the products most consulted in a store; or if the company manages car parks, it can automatically detect the number of free spaces. And these are just a few examples of what image recognition can do in the field of video surveillance.
Video surveillance cameras can become gold mines if we equip them with visual recognition technologies. They can serve multiple functions, for both public and private organizations, which to date still underestimate them too much. With both cameras already set up, data already available, and a data capture system already in place, much of the way is already paved for the development of image recognition systems. Companies and public authorities: seize the opportunity!
(1) Commission Nationale de l’Informatique et des libertés (CNIL) Press release – 21 June 2012 – “Vidéosurveillance/vidéoprotection: best practices for more privacy-friendly systems”