Driving early warning method and device based on multi-modal data, medium and program product

The invention provides a driving early warning method and device based on multi-modal data, a medium and a program product, and relates to the technical field of driving early warning. According to the method, multiple groups of driving data are acquired through acquisition, and the driving data comprise front vehicle reference data, driving state data of a driver of the vehicle and driving behavior data of the vehicle; according to the multiple groups of driving data and a preset recall strategy, a corresponding alarm mode is determined; and an alarm is given according to determined alarm mode. By applying the embodiment of the invention, the alarm is given based on the reference data of the front vehicle, the driving state data of the driver of the vehicle and the driving behavior data of the vehicle, so that comprehensive consideration is realized during alarming, the accuracy of alarming is improved, the driver is warned in time, and the safety coefficient of the driver is improved. 本申请提供一种基于多模态数据的驾驶预警方法、