Pedestrian trajectory prediction method and device based on adversarial network, equipment and medium

The invention discloses a pedestrian trajectory prediction method and device based on an adversarial network, equipment and a medium, and the method comprises the steps: obtaining past historical trajectory information and real future trajectory information of a current pedestrian, and carrying out the coding processing of the historical trajectory information, and obtaining a pedestrian motion state of the current pedestrian; processing the pedestrian motion state by using a social attention mechanism to obtain weight information of the current pedestrian, and obtaining predicted future trajectory information of the current pedestrian by using the weight information; performing auxiliary mutual information function processing on the predicted future trajectory information and the real future trajectory information to generate a loss function, and performing training optimization on an adversarial network by using the loss function to obtain a trained and optimized adversarial network; and performing pedestri