2020年5月16日 · 本文介绍了自导网络（SGN）如何利用自上而下的结构结合多尺度信息，提高图像去噪效果。 SGN在保持高性能的同时，提升了内存和运行效率。

2019年10月27日 · To tackle this problem, we propose a self-guided network (SGN), which adopts a top-down self-guidance architecture to better exploit image multi-scale information. SGN directly generates multi-resolution inputs with the shuffling operation.

To tackle this problem, we propose a self-guided network (SGN), which adopts a top-down self-guidance architecture to better ex-ploit image multi-scale information. SGN directly gener-ates multi-resolution inputs with the shuffling operation.

The code for my ICCV paper: Self-Guided Network for Fast Image Denoising. Shuhang Gu, Yawei Li, Luc Van Gool, and Radu Timofte. The code is based on the EDSR project: https://github.com/thstkdgus35/EDSR-PyTorch More results of this project will come shortly!

middle: additive Gaussian noise + clean image. right: denoised image using trained SGN. You can download pre-trained models (also on ILSVRC2012 validation set, different mu and sigma) via this link. You may use following script to test it on your own data (noted that you need to modify dataset path): zero mean Gaussian noise.

To tackle this problem, we propose a self-guided network (SGN), which adopts a top-down self-guidance architecture to better exploit image multi-scale information. SGN directly generates multi-resolution inputs with the shuffling operation.

2019年11月21日 · 提出了一种快速且高效的内存网络SGN来处理图像去噪任务。 通过采用自顶向下的自导体系结构，SGN在降噪性能，速度和内存效率方面均优于当前算法。

2019年10月1日 · SGN directly take shuffled images as multi-resolution inputs, and adopts a topdown self-guidance strategy to better utilizing contextual information. More details of SGN can be found in Sec. 3.

To tackle this problem, we propose a self-guided network (SGN), which adopts a top-down self-guidance architecture to better exploit image multi-scale information. SGN directly generates multi-resolution inputs with the shuffling operation.

About The PyTorch implementation of SGN (ICCV 2019), modified from https://github.com/zhaoyuzhi/Self-Guided-Network-for-Fast-Image-Denoising