Pedestrian re-identification method based on attribute information assistance

The invention discloses a pedestrian re-identification method based on attribute information assistance. The pedestrian re-identification method comprises the following steps: S1, marking pedestrian attributes on an ID dimension; S2, constructing a pedestrian re-identification deep learning network model, comprising a backbone network, an ID branch network and an attribute branch network, whereinthe backbone network serves as the input of the ID branch network and the attribute branch network, the ID branch network outputs a pedestrian global feature vector and predicts a pedestrian ID, and the attribute branch network outputs an attribute feature vector and predicts a pedestrian attribute; S3, constructing a loss function; S4, preparing a training data set, and training the pedestrian re-identification deep learning model; S5, extracting features of a queried image; and S6, calculating the similarity between the features of the inquired image and the image features of a image library. According to the method