Our SR duck output.We implement three different kinds of methods, CNN, guided diffusion models, and transformer, to enhance the quality of low-resolution images. Our work enhances SRCNN, combines different diffusion sampling algorithms and guided methods, and refines the process of facial image restoration on CodeFormer.
Jiahe Huang
Rising senior student, DS-Eng @ Umich, ECE @ SJTU
I am a rising senior student interested in computer vision and artificial intelligence for science.