Unmanned aerial vehicle video image small target tracking method based on twin network

The invention provides an unmanned aerial vehicle video image small target tracking method based on a twin network, and the method comprises the steps: carrying out the feature extraction of a template region and a search region through employing an improved VGG16 network, obtaining the feature maps of the template region and the search region, and improving the expression capability of features; enabling the template region feature map to sequentially pass through a channel attention channel and a space attention channel to improve the effective feature proportion, and obtaining an enhanced template region feature map; inputting the enhanced template region feature map and the search region feature map into an RPN network, respectively performing convolution operation on a classification branch and a regression branch, and obtaining a classification response map score and a regression response map score of the target; and finally, generating position information of the target in the next frame according to t