Video target re-identification method based on multi-frame feature fusion

The invention discloses a video target re-identification method based on multi-frame feature fusion. The method comprises the following steps: obtaining multi-frame continuous images of the same target; classifying the images according to the orientation of a target; performing target feature extraction on all the images; performing feature fusion and pooling on the images in the same direction toobtain fusion features; identifying the orientation of a to-be-identified target, and extracting to-be-identified features of the to-be-identified target; taking the product of the similarity of theto-be-identified feature and the fusion feature of the corresponding orientation and the weight factor of the orientation as the final similarity; and if the maximum value of the final similarity is greater than a given threshold value, determining that the identification is successful, and outputting a target corresponding to the maximum value of the final similarity as a re-identification result, otherwise, determining t