How Vision AI takes critical infrastructure security to the next level
Tracking people or objects in crowded public areas is difficult and time-consuming. Think of stations, shopping centres or stadiums. It is therefore crucial to analyse camera images quickly and accurately, especially in the event of incidents, missing persons or suspicious situations. Yet operators and investigation services struggle to retrieve relevant images from hours of footage, find the right person and find out where this person has been.
It is time-consuming and sensitive, as something can quickly be overlooked. The quicker a person can be traced, the quicker action can be taken by operators or investigators. This can limit the major consequences of a (potential) incident. In addition, many people are concerned about their privacy, due to the role of facial recognition and surveillance in this. This social tension calls for smart, ethical solutions.
The need for technology that supports personnel to efficiently search within imagery is rapidly increasing. Examples include tracking down missing persons of persons suspected in crime and other incidents. Facial recognition is always a delicate issue here though, partly due to the requirements set out in the AVG. But what if persons are not recognised on biometric data but on the basis of external features such as clothing colour, backpack or height? These points form the basis of a technology called Cross Camera Tracking.
Cross Camera Tracking is made possible by Video AI Analytics. But what exactly is Video AI Analytics? Video AI Analytics is a technology that uses artificial intelligence (AI) to automatically analyse, interpret and filter relevant information from video footage. Instead of a human having to spend hours reviewing camera footage, Video AI Analytcis can do this much faster and more consistently. For example, it can detect objects or people, analyse behaviour, monitor areas, count and track people, and provide automatic alerts in case of suspicious situations.
Cross Camera Tracking searches for individuals based only on external features without requiring biometric data. Here, smart filters can be enabled for clothes or backpacks, for example. For example, if an operator or investigation service is looking for a person dressed in all black, this person can be tracked. In doing so, Cross Camera Tracking searches all available camera images for similar appearances. By linking the results found to the camera location, it is possible to gain insight into where a person is (has been) via a map. Walking routes are also revealed in this way. The technology can be used both forensically and in real-time. Operators can conduct a targeted search for a suspicious or missing person based on external features immediately after a report. Cross Camera Tracking not only saves valuable time, but also prevents privacy breaches.
Indeed, handling privacy is central to the development and application of Cross Camera Tracking. Thus, the technology does not use facial recognition or other biometrics. Instead, it uses non-unique features that can change quickly, such as clothing and accessories. Moreover, images are only stored temporarily, access to the technology is often strictly regulated and all actions are carefully logged. By working according to privacy guidelines and conducting DPIAs (Data Protection Impact Assessments), the requirements of the AVG are demonstrably met.
Cross Camera Tracking brings several practical benefits. Consider, for example, faster recovery of missing persons and speeding up investigations into theft or violent incidents. It also significantly reduces the workload of operators or investigators, as hours of footage can be searched in minutes.
The success of this technology lies in the balance between security and privacy. Transparency, clear rules and oversight are essential to maintain citizens’ trust. The future lies with efficient and effective technology, without mass surveillance.
Wondering what this technology can offer your organisation? Cross Camera Tracking is one of 35+ AI Analytics offered by the Vaidio platform. If you want to know more, check out the product pageor contact us at: info@vaibs.com – +31 (0)85 4873031.