Exploiting the Potential of Standard Convolutional Autoencoders for Image Restoration by Evolutionary Search

By Masanori Suganuma, Mete Ozay, Takayuki Okatani. 2018

92 citations recorded by the source corpus.

Researchers have applied deep neural networks to image restoration tasks, in which they proposed various network architectures, loss functions, and training methods. In particular, adversarial training, which is employed in recent studies, seems to be a key ingredient to success. In this paper, we show that simple convolutional autoencoders (CAEs) built upon only standard network components, i.e., convolutional layers and skip connections, can outperform the state-of-the-art methods which employ adversarial training and sophisticated loss functions. The secret is to employ an evolutionary algorithm to automatically search for good architectures. Training optimized CAEs by minimizing the $\ell_2$ loss between reconstructed images and their ground truths using the ADAM optimizer is all we need. Our experimental results show that this approach achieves 27.8 dB peak signal to noise ratio (PSNR) on the CelebA dataset and 40.4 dB on the SVHN dataset, compared to 22.8 dB and 33.0 dB provided by the former state-of-the-art methods, respectively.

Subfields: Computer Vision and Pattern Recognition, Media Technology

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

In Media Technology

Citations in the corpus

Memory-Efficient Hierarchical Neural Architecture Search for Image Denoising (2019, methodology): “Two more relevant works are E-CAE [32] and FALSR [3].”
A Survey on Neural Architecture Search (2019, background): “For instance, there are first attempts to automatically optimize the architecture of autoencoders (Suganuma et al., 2018), transformers (So et al., 2019) and graph CNNs (Gao et al., 2019).”
A genetic programming approach to designing convolutional neural network architectures (2017, methodology): “The architecture search method explained in this paper has applied to automatically construct the autoencoder typed CNNs for the image denoising and inpainting tasks [Suganuma et al., 2018].”
Best Practices for Scientific Research on Neural Architecture Search (2019, background): “We note that application papers in NAS have already started tackling non-standard applications, such as image restoration (Suganuma et al., 2018), semantic segmentation (Chen et al., 2018; Nekrasov et al., 2018; Liu et al., 2019a), disparity estimation (Saikia et al., 2019), machine translation (So…”
CLEARER: Multi-Scale Neural Architecture Search for Image Restoration (2020, background): “In details, BM3D [4], WNNM [8], RED [27], MemNet [40], NLRN [19], DuRN-P [24], N3Net [31], HiNAS [46], and E-CAE [38].”
Techniques for Automated Machine Learning (2019, methodology): “Recently, AutoDL has been developed to deal with segmentation [58; 91], graph-CNN [29; 100], GAN [33], and autoencoders [80].”
Efficient Residual Dense Block Search for Image Super-Resolution (2019, methodology): “This success has been reproduced in many other computer vision tasks, e.g. segmentation (Liu et al. 2019) and image restoration (Suganuma, Ozay, and Okatani 2018).”
A Survey on Evolutionary Neural Architecture Search (2020, methodology): “Generally, some ENAS methods only take the encoder into evolution [25], [42] because the decoder part is symmetric with the encoder and can be derived accordingly.”
NAS-DIP: Learning Deep Image Prior with Neural Architecture Search (2020, methodology): “NAS algorithms have been successfully applied to various tasks, including image classification [53, 61, 63, 64, 89, 90], semantic segmentation [14,51,58], object detection [24], image restoration [17,72], and image generation [26].”