Biomedical Image Analysis Project
P1: Age domain translation using diffusion models
Diffusion models are a new type of generative model that can be used for various tasks such as image translation or segmentation. The aim of this project is to obtain a conditioned image (e.g., a photo of oneself) to dynamically change one's age so that one appears younger (e.g., as a baby) or older (e.g., as a retired, old, happy person). Therefore, first set everything up on a lab workstation to run a pre-trained stable diffusion model. In the second step, use a facial aging dataset, such as the FFHQ aging dataset, to fine-tune your model to produce artificially rejuvenated or aged faces.
Relevant information: Review of GAN approaches: https://ieeexplore.ieee.org/ielx7/6287639/9668973/09729822.pdf?tag=1 Rombach et al. - “High-Resolution Image Synthesis with Latent Diffusion Models” https://github.com/CompVis/stable-diffusion https://github.com/royorel/FFHQ-Aging-Dataset