Dear everyone,
This Saturday we are going to talk the paper "Meta-SR: A Magnification-Arbitrary Network for Super-Resolution". Why this article? Firstly, it is full of fun. The authors propose a novel method called Meta-SR to firstly solve super-resolution of arbitrary scale factor with a single model. The method mainly uses meta-learning to achieve the goal of "arbitrary scale". Secondly, I think it's a good material for learning how to apply meta-learning algorithms to specific problems.
All are welcome!
Best wishes and Happy Children's Day!:)
Biao Zhang