You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
------------------ 原始邮件 ------------------
发件人: "CalayZhou/MBNet" ***@***.***>;
发送时间: 2022年6月3日(星期五) 晚上10:51
***@***.***>;
***@***.******@***.***>;
主题: Re: [CalayZhou/MBNet] |w|的校正方式? (Issue #52)
您好,|w|的值反应在wr与wt上,而wr与wt会对最终的置信度施加权重,所以在分类损失反向传播时时会对|w|进行优化;不过MBNet的光照加权设计得比较粗糙,有挺大的改进空间~
—
Reply to this email directly, view it on GitHub, or unsubscribe.
You are receiving this because you authored the thread.Message ID: ***@***.***>
作者你好,我最近在阅读跨模态行人检测的论文,我想问一下文章中提到的|w|以什么方式来校正呢?
我看见论文中|w|也是用卷积和连接层预测的,wd和wn两个指标有损失函数校正,那么|w|怎么使它预测更准确呢?
The text was updated successfully, but these errors were encountered: