A trajectory prediction method, apparatus, device, medium and computer program product

Abstract

The application provides a trajectory prediction method and device, equipment, medium and computer program product, and relates to the technical field of artificial intelligence, and comprises the following steps: obtaining fusion trajectory features according to position information and time information of a target user at each check-in point in historical trajectory data; performing feature embedding on a spatial interval and a time interval between a target check-in point and a corresponding first adjacent check-in point in the plurality of check-in points to obtain a spatio-temporal correlation measurement feature of the target check-in point; extracting a historical trajectory correlation feature of the target check-in point according to the fusion trajectory feature, the spatio-temporal correlation measurement feature, the position information and the time information of each target check-in point, generating a dynamic correlation feature of a predicted trajectory, obtaining a prediction feature, and then obtaining a predicted trajectory corresponding to the target user. The spatio-temporal correlation degree feature is constructed by the spatial interval and the time interval between the check-in points, which can comprehensively capture the dynamic correlation of the pedestrian trajectory and enhance the accuracy and efficiency of the trajectory feature extraction.

Images