PaddlePaddle / PaddleClas

A treasure chest for visual classification and recognition powered by PaddlePaddle
Apache License 2.0
5.44k stars 1.17k forks source link

HSresnet #501

Closed wangsun1996 closed 2 years ago

wangsun1996 commented 3 years ago

Hello,in your paper, you propose that: Unlike Res2Net[9], we avoid producing the abundant and even redundant information contained in the feature maps and the network can learn richer feature representation.请问这个如何理解呢?从哪个角度说明新的HS网络能够避免提取冗余信息呢?感谢

cuicheng01 commented 3 years ago

Hi,您好,在HS-block中,有很多split操作(层级的split),部分通道split后不做任何需要参数和计算的操作,这样就避免了产生冗余的信息,而HS-block较Res2Net中的block更为复杂,这样更多的参数就学习了图像更多非冗余的特征。

paddle-bot-old[bot] commented 2 years ago

Since this issue has not been updated for more than three months, it will be closed, if it is not solved or there is a follow-up one, please reopen it at any time and we will continue to follow up. It is recommended to pull and try the latest code first. 由于该问题超过三个月未更新,将会被关闭,若问题未解决或有后续问题,请随时重新打开(建议先拉取最新代码进行尝试),我们会继续跟进。