search

greatlog / UnpairedSR

This is an offical implementation of the CVPR2022's paper [Learning the Degradation Distribution for Blind Image Super-Resolution](https://arxiv.org/abs/2203.04962)
165 stars 12 forks source link

请问一下作者ratios: [50, 200]是什么作用？ #11

Closed Fangzhenxuan closed 2 years ago

Fangzhenxuan commented 2 years ago

https://github.com/greatlog/UnpairedSR/blob/b6d0c05ad327f8aa4118aab852a49d94e0b80ac7/codes/config/PDM-SR/options/train/percep/2020Track2.yml#L16

greatlog commented 2 years ago

作用有二

  1. 调节src图片数据数量和tgt图片数量不相等的问题，详见 #12

  2. 将数据集图片重复多次，以增多一个epoch内的迭代次数，从而可以减少整个训练周期的epoch数，从而减少每个epoch开始前的等待时间和dataset中drop laset的影响。