[GSoC '19 OpenCV] Learning-based Super Resolution
This is the result of my 2019 GSoC project with OpenCV. My task was to add a learning-based super-resolution module to OpenCV, which is called 'dnn_superres'. I implemented two neural networks:
EDSR: http://openaccess.thecvf.com/content_...
FSRCNN: http://personal.ie.cuhk.edu.hk/~ccloy...
EDSR is a very high-performance network and FSRCNN is a very fast network.
The module allows you to very easily scale up images & video by a factor of 2, 3 or 4.
Official link to my project: https://summerofcode.withgoogle.com/p...
Link to project result: https://gist.github.com/Saafke/8e5b58...
My mentors during this project were Vladimir Tyan and Yida Wang.