关于setting2下的训练和测试问题

greatlog / DAN

This is an official implementation of Unfolding the Alternating Optimization for Blind Super Resolution

231 stars 41 forks source link

Closed leeziqiang closed 2 years ago

leeziqiang commented 2 years ago

你好，想问下setting2中train和test的yaml配置中input_pare=15,而在setting1中用10？

greatlog commented 2 years ago

setting2中的kernel相对更复杂，属于anisotropic, 我们推测需要更长的向量来进行编码。

wwlCape commented 2 years ago

@leeziqiang 朋友你训练x2的时候会出现不收敛的问题么？

leeziqiang commented 2 years ago

这倒没有