From ba23738515d921bb3c2b22dd9c2c8e00eb88cf9e Mon Sep 17 00:00:00 2001 From: Wayne Hung Date: Tue, 11 Sep 2018 14:21:41 +0200 Subject: [PATCH] Update README.md --- README.md | 21 ++++++++++++++++++++- 1 file changed, 20 insertions(+), 1 deletion(-) diff --git a/README.md b/README.md index 53f2deb..e6e3103 100644 --- a/README.md +++ b/README.md @@ -1,6 +1,25 @@ -# "Learning To Blend Photos," ECCV 2018 +# Learning To Blend Photos, ECCV 2018 +## Code/Data will be released soon + +This repo demonstrate the following paper: + +[Learning to Blend Photos](http://openaccess.thecvf.com/content_ECCV_2018/html/Wei-Chih_Hung_Learning_to_Blend_ECCV_2018_paper.html)
+[Wei-Chih Hung](https://hfslyc.github.io/), [Jianming Zhang](https://jimmie33.github.io/), [Xiaohui Shen](http://users.eecs.northwestern.edu/~xsh835/), [Zhe Lin](https://research.adobe.com/person/zhe-lin/), [Joon-Young Lee](https://joonyoung-cv.github.io/), and [Ming-Hsuan Yang](http://faculty.ucmerced.edu/mhyang/)
+Proceedings of the European Conference on Computer Vision (ECCV), 2018. + +Contact: Wei-Chih Hung (whung8 at ucmerced dot edu) + +# Introduction + +In this work, we aim to automate the photo blending process through deep reinforcement learning. We focus on a specific and popular photo blending style - Double Exposure. The image below shows some example results of our method: + +![](images/teaser.png) + +The figure below shows the overview of our system. The inputs of our method are two photos: foreground and background. We first train a quality network to evaluate the aesthetics quality of blending photos with human preference annotation on random blending photos. Then a deep reinforcement learning based agent is trained to optimize the parameter for the background alignment and photometric adjustment. Using the predicted parameters, the blending engine renders the final blending photo. + +![](images/framework.png) Please cite our paper if you find it useful for your research.