复现不出论文精度

[zhending111]

您好，请问是还有其他什么训练策略么？训练了几次均复现不出论文中的精度，S measure 连69都没超过

[dragonlee258079]

你好，我的训练策略都在train.py代码里了，至于复现不出精度，你可以检查一下环境是否一致，下载的训练数据组合是否正确

[zhending111]

您好，那个syn总共有几个呢?

[zhending111]

我就好像有两个syn，一个是add_navie另一个是add_navie_reverse2

[dragonlee258079]

synthesis strategy 是 follow CADC（Summarize and search: Learning consensus-aware dynamic convolution
for co-saliency detection） 方法的，合成策略就是正向合成和反向合成，一共是两个

[zhending111]

synthesis strategy 是 follow CADC（Summarize and search: Learning consensus-aware dynamic convolution for co-saliency detection） 方法的，合成策略就是正向合成和反向合成，一共是两个

好的，我去检查一下环境吧，训练数据应该是没问题，感谢及时回复

[ZhengPeng7]

Hi, @dragonlee258079, so, what is the performance of DMT without the data synthesis strategy in CADC? Since most CoSOD methods didn't use that, the performance without it can be instrumental for comparison. Thanks!

[dragonlee258079]

@ZhengPeng7 Thank you for your suggestion. I will provide the performance without the data synthesis strategy soon.