METHOD AND SYSTEM FOR AUTOMATED CLASSIFICATION OF VEHICLE DAMAGES Russian patent published in 2024 - IPC G06N20/00 G06Q40/08 G06Q30/02 G07C5/00 

FIELD: data processing.

SUBSTANCE: invention relates to a method and a system for classifying vehicle damage using machine learning algorithms. In the solution, telemetry data are obtained from the telematics device installed on the vehicle; generating a data array based on the obtained data; determining at least one impact of the vehicle using a first machine learning model which processes an array of telematic data, which detects anomalies in the readings of the telematics device; determining the configuration of the vehicle impact and the type of loss based on the obtained data array using a second machine learning model trained on a sample of data on the configuration of impacts of the vehicle, the state of movement of the vehicle and the vehicle parameters; determining a list of damaged parts taking into account hidden damages by configuration of detected impacts using a third machine learning model, trained on a sample of data on losses, taking into account telematic information in case of road accidents; and classifying vehicle damage based on a certain list of damaged parts.

EFFECT: high accuracy of determining vehicle damage.

13 cl, 9 dwg