search

sunny2109 / SAFMN

[ICCV 2023] Spatially-Adaptive Feature Modulation for Efficient Image Super-Resolution; runner-up method for the model complexity track in NTIRE2023 Efficient SR challenge
258 stars 14 forks source link

Uncertainty-based loss #49

Closed muslll closed 3 months ago

muslll commented 3 months ago

Hi, thanks for your research. On the AIS 2024 submission, it is said that "the proposed SAFMN++ is trained by minimizing a combination of the uncertainty-based MSE loss and FFT-based L1 loss ". However, I couldn't find those in this repository. Do you plan on releasing it as well? Thanks in advance.

sunny2109 commented 3 months ago

Hi, sorry for such a late reply. The implementation of uncertainty loss can be found in the featup](https://github.com/mhamilton723/FeatUp) repo.

muslll commented 3 months ago

Thanks!