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

训练的时候输入不是640的话,预测不出来框 #8

Closed
jiarenyf opened this issue Mar 28, 2021 · 8 comments
Closed

训练的时候输入不是640的话,预测不出来框 #8

jiarenyf opened this issue Mar 28, 2021 · 8 comments
Labels
question Further information is requested Stale

Comments

@jiarenyf
Copy link

发现训练的时候输入如果不是640大小的话,训出来的模型预测不出框。只有输入大小是640训练的,才能在预测的时候出现框,这是为什么呢?麻烦作者看下,谢谢。

@jiarenyf jiarenyf added the question Further information is requested label Mar 28, 2021
@hukaixuan19970627
Copy link
Owner

是demo文件还是自制的数据集出现这种情况呢?
(白天知乎找我的话我回复会及时一点)

@jiarenyf
Copy link
Author

是demo文件还是自制的数据集出现这种情况呢?

是自己的数据集

@hukaixuan19970627
Copy link
Owner

预测不出框,调低conf-thresh

@Allenstin
Copy link

调低那些阈值都没用,训练时候,排在后面的指标全是0。

@hukaixuan19970627
Copy link
Owner

[
epoch/epochs-1 , mem , loss_obx , loss_obj , loss_cls ,loss_angle ,
total_loss , targets_num , imgs.shape, precision , recall , mAP@.5 ,
mAP@.5-.95, val_loss_box, val_loss_obj, val_loss_cls, val_loss_angle
]
由于删除了在线评估程序,所以后续8个评估数据都是0是正常现象,旋转目标的评估方式比较特别,我另外写了一个evaluation.py。

@jiarenyf
Copy link
Author

[
epoch/epochs-1 , mem , loss_obx , loss_obj , loss_cls ,loss_angle ,
total_loss , targets_num , imgs.shape, precision , recall , mAP@.5 ,
mAP@.5-.95, val_loss_box, val_loss_obj, val_loss_cls, val_loss_angle
]
由于删除了在线评估程序,所以后续8个评估数据都是0是正常现象,旋转目标的评估方式比较特别,我另外写了一个evaluation.py。

我把test.py里面的test函数补全了,用poly_iou计算pred与target的iou进行匹配,在输入640大小的时候,是可以输出p/r/map这些的,指标可以训的比较高。但是用1024/1536这些大小输入的话,p/r/map指标都是0。。。

@hukaixuan19970627
Copy link
Owner

emmm, 不清楚,能否把你的数据集情况发过来看看。
另外预测不出框,是置信度调到0.05也不出框吗?如果是这样的话我去检查一下代码问题

@github-actions
Copy link

This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
question Further information is requested Stale
Projects
None yet
Development

No branches or pull requests

3 participants