How SophyAI® works (A.I. & Robotic Platform)
The SophyAI’s neural network can “classify and interpret” if properly trained, a scenario caught from a normal video surveillance camera.
The streaming video, coming from the cameras, are sent to the neural network that, through specific algorithms, identifies particular “objects” of the scene.
In addition to this, a specific geolocation module can identify the correct spatial position of the observed “objects” and transform it into geographic coordinates, representable on any map.
The third component of SophyAI, the workflow module, takes care of implement predefined “actions” in relation to the behavior of the classified “objects”. This is a very important function of the system, in fact it allows SophyAI to relate to other systems providing information and commands
Quantity, direction, velocity, and state of “objects” are processed by the workflow module to send information and alerts to operators.
SophyAI can also be interconnected with collaborative elements (IOT) present in the observed area, both to improve the understanding of the scene and send automatic commands as: close and open gates, smart parking management, activate fire-fighting systems, change traffic light timing, etc.
Smart Traffic by SophyAI®
Video surveillance cameras can allow SophyAI to interpret and analyze the city’s traffic condition and provide, in addition to useful information on its density by areas, timely indications of what happens in the traffic dynamics. The neural network, as already explained, can define the spatial conditions of what has been classified, in sense of directions and speeds. This information, with the appropriated algorithms, can be used for interpreting the traffic status (vehicles and persons) in various manners.
For example, in many cases the traffic lights timing is not appropriate to the real traffic conditions, SophyAI may decide to change the traffic lights synchronization to perform a better traffic management, giving priority to the road line with more vehicles. If a vehicle stops, improperly or for failures, in the driving lane, the neural network can be instructed to analyze the state of vehicles movement and interpret whether it can generate danger or not. In the event of traffic, SophyAI interprets slowdowns or stops as a normal condition, but if the neural network does not detect traffic and recognizes that a vehicle is stationary in a predetermined area (driving lane) for a certain period (higher than that due for Example to a normal parking), can generate appropriate alarms. The workflow module immediately can send an alarm and eventually act on the traffic light to prevent accidents.
SophyAI can carry out a precise counting of the vehicles passing and classing them in relation to the type of the vehicle itself. In the same way SophyAI can count people Beyond the value of statistical counting, this function can be very useful if interfaced with collaborative sensors (IOT) as traffic light network or smart devices used by municipal police.