We present a neural architecture search (NAS) technique to enhance the performance of unsupervised image de-noising, in-painting and super-resolution under the recently proposed Deep Image Prior (DIP). We show that evolutionary search can automatically optimize the encoder-decoder (E-D) structure and meta-parameters of the DIP network, which serves as a content-specific prior to regularize these single image restoration tasks. Our binary representation encodes the design space for an asymmetric E-D network that typically converges to yield a content-specific DIP within 10-20 generations using a population size of 500. The optimized architectures consistently improve upon the visual quality of classical DIP for a diverse range of photographic and artistic content.
@article{arxiv.2001.04776,
title = {Neural Architecture Search for Deep Image Prior},
author = {Kary Ho and Andrew Gilbert and Hailin Jin and John Collomosse},
journal= {arXiv preprint arXiv:2001.04776},
year = {2020}
}