chunmeifeng / SANet

【IEEE TNNLS 2023】Exploring Separable Attention for Multi-Contrast MR Image Super-Resolution
43 stars 6 forks source link

论文复现 #2

Open Itsanewday opened 1 year ago

Itsanewday commented 1 year ago

作者您好,首先感谢公开代码。我想问下论文里与其他公开的论文结果做了一些比较,这些其他论文的结果是您自己复现的吗?比如SMORE,它是针对每个样本单独训练一个超分辨模型,请问这个该怎么实现?谢谢

chunmeifeng commented 1 year ago

是需要自己复现的

2023年1月5日 上午10:05,Itsanewday @.***> 写道:

作者您好,首先感谢公开代码。我想问下论文里与其他公开的论文结果做了一些比较,这些其他论文的结果是您自己复现的吗?比如SMORE,它是针对每个样本单独训练一个超分辨模型,请问这个该怎么实现?谢谢

— Reply to this email directly, view it on GitHub https://github.com/chunmeifeng/SANet/issues/2, or unsubscribe https://github.com/notifications/unsubscribe-auth/AR75XNYD6EZDL6ULO55Q6V3WQYT55ANCNFSM6AAAAAATRNSIGM. You are receiving this because you are subscribed to this thread.

Itsanewday commented 1 year ago

是需要自己复现的 2023年1月5日 上午10:05,Itsanewday @.***> 写道: 作者您好,首先感谢公开代码。我想问下论文里与其他公开的论文结果做了一些比较,这些其他论文的结果是您自己复现的吗?比如SMORE,它是针对每个样本单独训练一个超分辨模型,请问这个该怎么实现?谢谢 — Reply to this email directly, view it on GitHub <#2>, or unsubscribe https://github.com/notifications/unsubscribe-auth/AR75XNYD6EZDL6ULO55Q6V3WQYT55ANCNFSM6AAAAAATRNSIGM. You are receiving this because you are subscribed to this thread.

是需要自己复现的 2023年1月5日 上午10:05,Itsanewday @.***> 写道: 作者您好,首先感谢公开代码。我想问下论文里与其他公开的论文结果做了一些比较,这些其他论文的结果是您自己复现的吗?比如SMORE,它是针对每个样本单独训练一个超分辨模型,请问这个该怎么实现?谢谢 — Reply to this email directly, view it on GitHub <#2>, or unsubscribe https://github.com/notifications/unsubscribe-auth/AR75XNYD6EZDL6ULO55Q6V3WQYT55ANCNFSM6AAAAAATRNSIGM. You are receiving this because you are subscribed to this thread.

感谢回复,我想知道我关于SMORE的理解是否和你一致,因为如果对于每个样本SMORE都要微调模型,那么它的效率就很低,难以应用。还是说SMORE对于一个类似的数据集,比如全都是脑部的,仅需要一个模型。最近我在做核磁超分辨相关的研究,还望不吝赐教。