Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

训练HRSC2016数据集 #48

Open
wulele2 opened this issue Jul 4, 2021 · 13 comments
Open

训练HRSC2016数据集 #48

wulele2 opened this issue Jul 4, 2021 · 13 comments

Comments

@wulele2
Copy link

wulele2 commented Jul 4, 2021

大佬您好,我训练HRSC2016数据集最终train loss和val loss维持在4左右就不变了,测试了demo发现预测效果也不好。能指点一二吗?谢谢了

@chenpeng68
Copy link

大佬,我训练自己的数据集之后怎么评估

@jacksparrow-firstmate
Copy link

别想了,这种热图关键点检测在HRSC上面用不了的

@xing-bing
Copy link

别想了,这种热图关键点检测在HRSC上面用不了的

老铁,为什么这在HRSC上面用不了啊

@xing-bing
Copy link

大佬您好,我训练HRSC2016数据集最终train loss和val loss维持在4左右就不变了,测试了demo发现预测效果也不好。能指点一二吗?谢谢了

请问你的HRSC2016数据集跑成功了吗

@wulele2
Copy link
Author

wulele2 commented May 26, 2022

成功了,map80多

@xing-bing
Copy link

成功了,map80多

挺高啊,我一直没有跑成功,我训练完之后输出的角度值一直都是负的,也不知道错在哪了

@wulele2
Copy link
Author

wulele2 commented May 26, 2022

仔细检查下hrsc2016的标注格式,将其转成[cx,cy,long,short,theta], theta属于[0,pi),然后可视化看下转的对不对。我损失下不去就是因为数据label转换错了,希望帮到你

@xing-bing
Copy link

HRSC2016,那个角度是用弧度表示的, 有正有负

@xing-bing
Copy link

仔细检查下hrsc2016的标注格式,将其转成[cx,cy,long,short,theta], theta属于[0,pi),然后可视化看下转的对不对。我损失下不去就是因为数据label转换错了,希望帮到你

能不能让我看看你的数据是怎么处理的老铁

@wulele2
Copy link
Author

wulele2 commented May 26, 2022

这两天我上传到GitHub吧

@xing-bing
Copy link

这两天我上传到GitHub吧

好的非常感谢,这个错误找好久了

@xing-bing
Copy link

这两天我上传到GitHub吧

兄弟程序搞完了吗

@wulele2
Copy link
Author

wulele2 commented May 28, 2022

今天上传,可能有点儿凌乱,不过应该能跑

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

4 participants