VEHICLE SYSTEM OF A VEHICLE FOR DETECTING AND VALIDATING AN EVENT USING A DEEP LEARNING MODEL

摘要

The invention relates to a vehicle system (1) of a vehicle (2) configured to detect an event (E) and to broadcast said event (E) using a decentralized environmental notification message (DENM), wherein said vehicle system (1) comprises:- at least one camera sensor (10) configured to capture images (I1) of an environment of said vehicle (2),- an electronic control unit (11) configured to :- detect an event (E) using a primary deep learning model (M1) based on said images (I1),- apply an predictability level (A) on said event (E), said predictability level (A) being generated by said primary deep learning model (M1),- transmit said event (E) to a telematic control unit (12) if its predictability level (A) is above a defined level (L1),- said telematic control unit (12) configured to :- receive said event (E) from said electronic control unit (10) and broadcast a related decentralized environmental notification message (DENM) via a vehicle to vehicle communication (V2V) and/or a vehicle to infrastructure communication (V2I),- transmit at least one image (I1) and data details (D) of said event (E) to a server (3),- receive a primary validation information (30) of said event (E) from said server (3), said primary validation information (30) being generated by a secondary deep learning model (M2), and cancel the broadcasting of said decentralized environmental notification message (DENM) if said event (E) is not validated,- if said event (E) is validated, receive an updated instance (M3) of said primary deep learning model (M1) from said server (3) and transmit it to said primary electronic control unit (10) for updating said primary deep learning model (M1).