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

多尺度测试(multi-scale test)中的问题 #1044

Open
automatic0313 opened this issue Jul 2, 2024 · 5 comments
Open

多尺度测试(multi-scale test)中的问题 #1044

automatic0313 opened this issue Jul 2, 2024 · 5 comments

Comments

@automatic0313
Copy link

What's the feature?

您好,我能顺利的使用multi-scale training的流程,但在测试阶段
test_pipeline = [
dict(type='LoadImageFromFile'),
dict(
type='MultiScaleFlipAug',
img_scale=[(1024, 1024),(1440,900),(900,1440)]
flip=False,
transforms=[
dict(type='RResize',multiscale_mode='value'),
dict(type='Normalize', **img_norm_cfg),
dict(type='Pad', size_divisor=32),
dict(type='DefaultFormatBundle'),
dict(type='Collect', keys=['img'])
])
]

Any other context?

No response

@automatic0313
Copy link
Author

这样的配置无法跑通,请问我该如何修改呢

@toxinv
Copy link

toxinv commented Aug 18, 2024

源码里面貌似没给具体的实现

@yangxue0827
Copy link
Collaborator

@automatic0313
Copy link
Author

您可能没理解我的意思,mmdetection中多尺度训练不是每隔一定的iterations就随机一个图像尺寸吗,推理时是多个尺度并行推理,然后NoC合框,您这回复并不是多尺度推理的,而是多尺度数据集训练推理的

@yangxue0827
Copy link
Collaborator

mmrotate里常用的多尺度训练和测试就是离线的,其实是离线多尺度裁剪,这是因为DOTA有裁剪图像的操作,所以实验下来离线的会比在线的有效。mmdet那种在线的方式我没试过支不支持。

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

3 participants