Partially supported by Microsoft Research through a Data Science Award
We are living in an era of massive exodus and delicate security issues. These exodus creates security problems that countries have to deal with in order to provide aid as people cross their territories, and particularly, to be sure that criminal organizations do not take advantage of their situation.
We believe that technology, particularly data management and visualization, can provide tools for helping controlling this complex situation. Thus CONCERNS aims at:
- Harvesting data collections of on-line observations performed in urban spaces.
- Analysing and visualizing these collections at different granularities levels in 3D.
- Performing crowd behaviour simulations and detecting exceptional situations.
The challenge we are addressing is the one of combining individuals’ locations (like those provided by the Twitter or QQ services) with real-time graphic vision of the urban environment they are moving within (like Google Earth). For this we count on using real data produced by GPS, mobile and telephone networks, as well as static and mobile cameras to predict individual and crowd behaviour in presence of specific events.
The partners of the project will share their expertise and previous results on DRONES, data storage/analytics and visualization of crowds. The objective is thus to build a system (cf. figure below) that:
- Observe individuals within crowds as they evolve in specific urban spaces, create data collections and transmit data on-line using DRONES and Model2Roo toolkit for storing data on NoSQL stores.
- Detect suspicious behaviours and subjects within the crowd that might not belong to it (e.g. mafia, people dealers), by applying data analytics techniques, and particularly, relations inference within graphs.
- Simulate crowd behaviour in the presence of specific events that might create disturbances, or isolate, wear or disabled elements of the crowd using GPUs, HPC infrastructures and real data.