How to convert UNet to decode_head? #2178
Unanswered
WandernForte
asked this question in
Q&A
Replies: 1 comment
-
Hi, UNet is an encoder-decoder architecture which is a widely used backbone for segmentation model. In MMSegmentation, its UNet decoder head uses various feature maps generated and concatenated from different stages(aka downsampling and upsampling blocks). It could be FCN, DeepLabV3 and PSPNet. For example, the deeplabv3 as decoder head in MMSegmentation UNet: So if you want to use other decoder heads, you just need to modify decoder head in your config file and make sure the input dimension match up with its feature map requirements. Best, |
Beta Was this translation helpful? Give feedback.
0 replies
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
-
it is an implementation from segmentation_models.pytorch
it runs very slowly, and i don;t know whether it works or not...
Beta Was this translation helpful? Give feedback.
All reactions