⚠️⚠️⚠️Repo Under Construction⚠️⚠️⚠️

FourierMamba

Unofficial Implementation of FourierMamba Block from the FourierMamba paper, for image de-raining using State Space Models.

Usage

1 FourierMamba Block

import torch
from fourierMamba import FourierMamba



model = FourierMamba(inp_channels=3, out_channels=3, dim=48)
x = torch.randn(2, 3, 128, 128)
out = model(x)

Todo

• reduce time complexity on scanning.
• add analysis on scanning methods.
• adding a training and evaluation script.
• Fix shape mismatch in FourierMama class.

References

1. aSleepyTree/FreqMamba

Citations

@misc{@article{Li2024FourierMambaFL,
  title={FourierMamba: Fourier Learning Integration with State Space Models for Image Deraining},
  author={Dong Li and Yidi Liu and Xueyang Fu and Senyan Xu and Zheng-Jun Zha},
  journal={ArXiv},
  year={2024},
  volume={abs/2405.19450}
}