Age-dependent face diversification via latent space analysis

Facial age transformation methods can change facial appearance according to the target age. However, most existing do not consider that people get older with different attribute changes (e.g., wrinkles, hair volume, and face shape) depending on their circumstances environment. Diversifying such age-dependent attributes while preserving a person’s identity is crucial broaden applications of transformation. In addition, accuracy childhood limited due dataset bias. To solve these problems, we propose an method based latent space analysis StyleGAN. Our obtains diverse age-transformed images by randomly manipulating in space. so, analyze perturb channels affecting attributes. We then optimize perturbed code refine output image. also present unsupervised approach for improving childhood. assumption cannot sufficiently move toward desired direction. extrapolate estimated path iteratively update along extrapolated until image reaches Quantitative qualitative comparisons show our improves diversity preserves identity. more accurately perform