Target tracking method based on twin network and attention

According to the target tracking method based on the twin network and the attention, the initial frame of the target is given, and the position and the size of the target in the current frame can be automatically and accurately positioned in a subsequent video. Extracting feature information of the last three blocks based on the modified residual network; enhancing the feature information of the target based on a cross attention module, and inhibiting the influence of a background and an interferent; the enhanced feature map is sent into the improved head network, and cross-correlation operation is carried out on each block of features to obtain classification, centrality and regression results; and according to the result, the classification score and the centrality score are multiplied, a more accurate classification result is output, and the accurate position and size of the target in the current frame are obtained. According to the invention, by combining the cross attention module and improving the head