“In the field of computer vision, data generation techniques based on generative adversarial networks and diffusion models have provided new impetus for the construction of image and video datasets, promoting technological development.”