Advanced image recognition helps find the owners of unattended luggage
Partners: European airport
An expensive and disruptive problem
Unattended luggage is a major issue for airport operators. Although bags are often left by accident, the potential security risk means that if their owners can’t be tracked down quickly, the entire terminal may have to be evacuated and closed. Most unattended bags are false alarms. But each instance must be taken seriously, so operators deploy large teams of security staff at great expense to make sure even costlier disruption doesn’t happen.
Taking advantage of an existing resource: CCTV
Part of the solution is already installed in modern airports: pervasive CCTV systems mean that passengers and their luggage alike are under constant recorded surveillance. However, reviewing this footage manually to reunite bags with their owners is a prohibitively expensive and time-consuming task. Seeking to explore how CCTV could be used to solve the unattended luggage problem, a major European airport teamed up with Hitachi in Europe and our Global Centre for Social Innovation to explore a new way of matching unattended luggage with its owners.
The questions we wanted to answer were simple. Who left the bag? And where is that person now? We started by understanding how current unattended bags were processed by airport security operations. Our user-centred design methodology makes sure we know the customer’s needs, KPIs and ways of working so that future solutions can be embedded appropriately into their operation.
Hitachi’s C2X middleware platform is a next-generation system that enables vehicles to
exchange information for safer, greener driving:
Communication in motion
Hitachi’s C2X middleware platform is a next-generation system that enables vehicles to exchange information for safer, greener driving:
Unique mix of AI technologies
We developed a solution that for the first time blended artificial intelligence technologies for both object detection and person tracking. To spot unattended luggage, the object-detection system uses algorithms to detect whether an object of a certain size, such as a bag, appears within a video frame. It then starts a timer that raises an alert if the object, in this case luggage, is left alone for a specific time period. The system then suggests the moment when the bag was left by its owner, which is confirmed by the security team. At this point, the team can decide whether to invoke Hitachi’s person-tracking technology to find the individual. This advanced system applies deep learning to physical attributes – for example the colour and style of hair and clothing – to accurately identify, locate and track the individual in real time. Unlike facial recognition, it doesn’t depend on people’s position relative to cameras or cause privacy issues.