FOCUS
Smart city: how to manage risks with Computer Vision
The expression Smart City is no longer an abstract concept, but the natural propensity of urban agglomerations for the intensive use of the same ‘smart’ technologies that have already made their way in the productive / business field and also in people’s everyday life. Certainly, the implementation of technologies such as Machine Learning, Computer Vision and NLP (Natural Language Processing) in the urban context brings with it higher level challenges, above all in consideration of the disruptive impact that these would have on the everyday life of millions of people. Moreover, the areas of application of a data-driven urban model would be extremely wide: from intelligent mobility, with (totally) autonomous driving cars, to environmental monitoring systems capable of informing citizens in real time and directing their behavior, not to mention smart crime prevention systems or IoT-based health ecosystems that connect patients, physicians and facilities.
The advantages of being smart
The paradigm shift is necessary not only to support and exploit the technological developments to the maximum by pointing them towards the wellbeing of citizens and administrations, but to manage the strong and continuous growth of urban agglomerations: according to the United Nations, in 2030, urban areas, including the so-called megacities, will host at least 60% of the world population, which requires new organizational and governance models to meet the needs of cities of tens of millions of people. The transition to a data-driven urban model thus brings with it new challenges and, above all, requires a review of Risk Management that takes the utmost account of the new role of “machines”, which can support the activities of citizens, make decisions and independently manage some activities that are now reserved for human beings, from driving on the road to managing critical infrastructures. If IT security is already central for a thousand different reasons, in the smart city paradigm, sensors, devices and connectivity services will be fundamental for most of the services dedicated to citizens: an interruption in service, a failure or a targeted data attack could have disastrous consequences, that have to be avoided at all costs.
The role of Computer Vision in a Smart City
Given that the smart city concept involves a review of risk management models, the technologies that enable it provide administrations with essential data with which to evaluate, manage and mitigate urban and territorial risks: floods, road risks, hydraulic risks, assessment of degradation, state of the infrastructures etc. Among these technologies, one cannot fail to mention Computer Vision, which can be defined as the ability of machines to replicate human vision both at the level of image acquisition and – above all – of interpretation of the image. Thanks to the clear progress in acquisition quality and processing speed by small devices that can be distributed in the city perimeter, Computer Vision is one of the technological foundations of the Smart City.
When we talk about Computer Vision we associate it with the concept of face recognition, a concept made famous by consumer devices like the smart phones of recent generations, where it is used for unlocking devices. The chronicle then speaks to us of advanced biometric recognition systems based on Computer Vision techniques, employed in some airports for immigration procedures, but you can use your imagination to realize how many applications a technology of this kind, strongly soaked with Artificial Intelligence, could have in the context of a Smart City: the first hypotheses that come to mind concern the prevention of criminal activities, but also the ability to identify missing persons by recognizing them from the face or from the clothes worn. All this can pose questions of regulatory compliance, but from a technical point of view the benefits for the community would be more than tangible.
Computer Vision and Machine Learning: taking action before the problem
The potential of Computer Vision, however, goes far beyond the ability to recognize objects and people, because through this technology smart cities can assess the movements of subjects in the urban context, as well as monitor the state of the infrastructure over time, detecting any surface damage or deeper defects that need action. Not only that: an image recognition system based on Machine Learning can identify and predict traffic jams, detect areas that are perpetually overcrowded and request decisions related to road conditions that simplify people’s lives, make travel easier and ensure maximum safety. All this adds up to the possibility of detecting accidents, calling for emergency assistance, reporting road traffic violations and stops, and irregular occupancy of parking areas.
Like all smart systems, there are no limits to practical applications, also because the data deriving from Computer Vision can be enriched with other information that comes from external systems: for example, information on traffic, on the position of public transport, on the local weather and its forecasts. It is from the wise mix of all these analytics that public administrations can establish an increasingly evolving city and, despite the considerable size, also on a human scale.