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

HRNet or LiteHRNet? #68

Open
uk9921 opened this issue Jan 5, 2022 · 4 comments
Open

HRNet or LiteHRNet? #68

uk9921 opened this issue Jan 5, 2022 · 4 comments

Comments

@uk9921
Copy link

uk9921 commented Jan 5, 2022

>>> from mmpose.models import HRNet

In the document-example, HRNet is written instead of LiteHRNet.
I'm not sure if it is written incorrectly or deliberately. Is it wrong?

@uk9921
Copy link
Author

uk9921 commented Jan 5, 2022

from mmpose.models import LiteHRNet
import torch
extra = dict(
            stem=dict(stem_channels=32, out_channels=32, expand_ratio=1),
            num_stages=3,
            stages_spec=dict(
                num_modules=(2, 4, 2),
                num_branches=(2, 3, 4),
                num_blocks=(2, 2, 2),
                module_type=('LITE', 'LITE', 'LITE'),
                with_fuse=(True, True, True),
                reduce_ratios=(8, 8, 8),
                num_channels=(
                    (40, 80),
                    (40, 80, 160),
                    (40, 80, 160, 320),
                )),
            with_head=True,
        )
self = LiteHRNet(extra, in_channels=3)
self.eval()
inputs = torch.rand(1, 1, 32, 32)
level_outputs = self.forward(inputs)

The above is my example, I‘m not sure whether these default parameters are set correctly.
Correct me if I am wrong.

@nsapkota417
Copy link

Hi, I have been using LiteHRNet for Semantic Segmentation. Your parameters look fine for Lite-HRNet-18. For Lite-HRNet-30, change num_modules to (3, 8, 3) and it should be good.

@sunjian2015
Copy link

Hi, I have been using LiteHRNet for Semantic Segmentation. Your parameters look fine for Lite-HRNet-18. For Lite-HRNet-30, change num_modules to (3, 8, 3) and it should be good.

Hi, have you compared with HRNet-w16 on Semantic Segmentation? In this paper, the mIoU is higher than HRNet-16, but i got the opposite result, i don't know why

@yangjianmao
Copy link

nice

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