论文标题
神经架构寻找深层图像先验
Neural Architecture Search for Deep Image Prior
论文作者
论文摘要
我们提出了一种神经体系结构搜索（NAS）技术，以增强在最近提出的深层图像先验（DIP）下，无监督图像脱落，镶嵌和超分辨率的性能。我们表明，进化搜索可以自动优化DIP网络的编码器 - 编码器（E-D）结构（E-D）结构和元参数，该结构在正规化这些单个图像恢复任务之前是特定于内容的。我们的二进制表示编码非对称E-E-网络的设计空间，该网络通常会收敛，以使用500人口大小在10-20代内产生特定于内容的倾角。优化的体系结构始终如一地改善了经典浸入的视觉质量，以供各种摄影和艺术内容范围。
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.
