Real Image Super Resolution Via Heterogeneous Model Ensemble using GP-NAS
Abstract
With advancement in deep neural network (DNN), recent state-of-the-art (SOTA) image superresolution (SR) methods have achieved impressive performance using deep residual network with dense skip connections. While these models perform well on benchmark dataset where low-resolution (LR) images are constructed from high-resolution (HR) references with known blur kernel, real image SR is more challenging when both images in the LR-HR pair are collected from real cameras. Based on existing dense residual networks, a Gaussian process based neural architecture search (GP-NAS) scheme is utilized to find candidate network architectures using a large search space by varying the number of dense residual blocks, the block size and the number of features. A suite of heterogeneous models with diverse network structure and hyperparameter are selected for model-ensemble to achieve outstanding performance in real image SR. The proposed method won the first place in all three tracks of the AIM 2020 Real Image Super-Resolution Challenge.
Cite
@article{arxiv.2009.01371,
title = {Real Image Super Resolution Via Heterogeneous Model Ensemble using GP-NAS},
author = {Zhihong Pan and Baopu Li and Teng Xi and Yanwen Fan and Gang Zhang and Jingtuo Liu and Junyu Han and Errui Ding},
journal= {arXiv preprint arXiv:2009.01371},
year = {2021}
}
Comments
This is a manuscript related to our algorithm that won the ECCV AIM 2020 Real Image Super-Resolution Challenge