1 paper
Haorui Song, Yong Du, Tianyi Xiang +3
Despite the demonstrated editing capacity in the latent space of a pretrained GAN model, inverting real-world images is stuck in a dilemma that the reconstruction cannot be faithfu…