Real Image Super Resolution Via Heterogeneous Model Ensemble using GP-NAS
This addresses the problem of real-world image enhancement for applications like photography and surveillance, but it is incremental as it builds on existing dense residual networks.
The paper tackled real image super-resolution where both low- and high-resolution images are from real cameras, using a Gaussian process-based neural architecture search to select heterogeneous models for ensemble, achieving first place in all tracks of the AIM 2020 challenge.
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.