Pedestrian detection method and system based on MPDNet model

The invention discloses a pedestrian detection method and system based on a metro scene pedestrian detection network MPDNet model. The method comprises the following steps: obtaining original data of a metro station, and processing the original data to form a MetroStation data set; inputting the training set of the data set into a PVT model to extract a sample image feature map, and inputting the sample image feature map into an ASFF-FPN feature pyramid network for feature fusion to obtain a feature map after feature fusion; sending the feature map after feature fusion to a RetinaNet head module, and improving a bounding box regression loss function GIoU loss to obtain the category and position information of the target; carrying out model training by adopting the steps to form an MPDNet model; and carrying out pedestrian detection by using an MPDNet model. According to the method, the PVT model is used in cooperation with the ASFF-FPN with the improved structure on the whole to carry out feature map output,