NAS-DIP: Learning Deep Image Prior with Neural Architecture Search

By Yunpeng Chen, Chen Gao, Esther Robb, Jia-Bin Huang. 2020

64 citations recorded by the source corpus.

Recent work has shown that the structure of deep convolutional neural networks can be used as a structured image prior for solving various inverse image restoration tasks. Instead of using hand-designed architectures, we propose to search for neural architectures that capture stronger image priors. Building upon a generic U-Net architecture, our core contribution lies in designing new search spaces for (1) an upsampling cell and (2) a pattern of cross-scale residual connections. We search for an improved network by leveraging an existing neural architecture search algorithm (using reinforcement learning with a recurrent neural network controller). We validate the effectiveness of our method via a wide variety of applications, including image restoration, dehazing, image-to-image translation, and matrix factorization. Extensive experimental results show that our algorithm performs favorably against state-of-the-art learning-free approaches and reaches competitive performance with existing learning-based methods in some cases.

Subfields: Computer Vision and Pattern Recognition, Model Architectures

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Computer Vision and Pattern Recognition

1. Yujiu Yang
2. Ming-Hsuan Yang
3. Weihao Xia
4. Yulun Zhang
5. Jing-Hao Xue
6. Luc Van Gool
7. Bolei Zhou
8. Lei Zhang

In Model Architectures

1. Bolei Zhou
2. Ming-Hsuan Yang
3. Jing-Hao Xue
4. Yulun Zhang
5. Yujiu Yang
6. Weihao Xia

Citations in the corpus

• GAN Inversion: A Survey (2021, background): “Some other studies [204], [205] propose to search for neural architectures that capture stronger image priors instead of using hand-designed architectures.”