Tamper detection method based on convolution vision Mama

The invention discloses a tampering detection method based on convolutional vision Mama, and the method comprises the steps: firstly processing a data set through employing a seam cutting algorithm, and reducing the width of a picture according to different seam cutting rates, thereby obtaining the data set; secondly, randomly cutting the processed original image to a size which is the same as that of the processed image with the corresponding seam cutting rate; and then inputting the data set into a network model based on convolutional vision Mama, calculating classification loss by using cross entropy, and updating network model parameters by using back propagation. And finally, inputting a to-be-detected image into the trained network model, and outputting a detection result that whether the image is subjected to seam cutting processing or not. According to the method, artifacts existing in the image can be captured more easily, and the detection accuracy can be effectively improved when low-seam-cutting-r