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

Difference between Unet-full-sup and Geo-aware unet #13

Open
Carina02 opened this issue Mar 30, 2024 · 2 comments
Open

Difference between Unet-full-sup and Geo-aware unet #13

Carina02 opened this issue Mar 30, 2024 · 2 comments

Comments

@Carina02
Copy link

Carina02 commented Mar 30, 2024

Hello
After checking the code I still can not find the difference between the unet model with and without geo-aware. They are both based on the CoreUnet model so I see no difference on the model structures.
Can you explain what is the main difference between these two?

@91097luke
Copy link
Owner

91097luke commented Apr 2, 2024

Hi,
The model architecture is indeed the same. The only difference is initial initialization weights.
CoreUnet has randomly initialized weights while the geo-aware model has pre-trained encoder weights (a multi-head classifier pre-training approach was used to predict the co-oridinates, capture time & majority kg class of S2 patches)

@Carina02
Copy link
Author

Carina02 commented Apr 9, 2024

Hi, The model architecture is indeed the same. The only difference is initial initialization weights. CoreUnet has randomly initialized weights while the geo-aware model has pre-trained encoder weights (a multi-head classifier pre-training approach was used to predict the co-oridinates, capture time & majority kg class of S2 patches)

Thank you

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

2 participants